Sensing device, data processing unit, and sensing system

Abstract

A sensing system 1 comprises a sensing device 10 and a data processing unit 20. The sensing device 10 encrypts a detection signal by inserting a randomly selected digital watermark and pseudo noise generated every time a detection signal is acquired, and transmits the encrypted signal to the outside. The next seed value is obtained by adding an index value corresponding to the current electronic watermark to the current seed value, and pseudo noise corresponding to the sum is generated. The data processing unit 20 removes the pseudo noise and the electronic watermark every time a detection signal is received, and decodes the signal. The next seed value is obtained by estimating the electronic watermark which is the current electronic watermark and has the highest correlation among all the electronic watermarks, and adding an index value corresponding to the estimated electronic watermark to the current seed value, and pseudo noise corresponding to the sum is generated. Thus, the sensing data to be transmitted can be protected by means of a noise masking encryption method.

Description

Sensing device, data processing unit and sensing system

[0001] The present invention relates to a sensing device that encrypts and wirelessly transmits a sensing signal, a receiving-side data processing unit, and a sensing system.

[0002] In recent years, with the spread of the Internet of Things (IoT), there has been an increasing demand for sensing systems that wirelessly transmit sensing data from sensors to edge computers (data processing units). When sensing data contains highly confidential information, security measures are generally implemented for wireless transmission. For example, when sensing personal information such as biosignals (electroencephalogram signals), power-saving compressed sensing systems have been proposed that transmit compressed biosignals to a data processing unit via a wearable sensor attached to the body (e.g., Patent Literature 1, Non-Patent Literatures 1-3). While compressed sensing not only saves power but also functions as a lightweight encryption method, several security vulnerabilities have been identified due to the linear transformation involved (Non-Patent Literature 4).

[0003] To compensate for this vulnerability, Non-Patent Document 4 proposes a technology that applies a digital watermark to a compressed biosignal (electrocardiogram signal) to enhance robustness when the signal is transmitted to the data processing unit. Non-Patent Document 4 proposes a method in which several types of digital watermark signals prepared in advance are shared and stored by the sensor side and the data processing unit side. By incorporating a digital watermark into the transmitted biosignal, the signal appears to an eavesdropper as containing a lot of noise, and this is considered to be one of the lightweight security methods.

[0004] Japanese Patent Application Laid-Open No. 2021-115429

[0005] D. Kanemoto, et al., “Compressed Sensing Framework Applying Independent Component Analysis after Undersampling for Reconstructing Electroencephalogram Signals,” IEICE Trans. Fundamentals, vol. E103-A, no. 12, pp. 1647-1654, Dec. 2020.Y. Okabe, et al., “Compressed Sensing EEG Measurement Technique with Normally Distributed Sampling Series,” IEICE Trans. Fundamentals, vol.E105-A, no. 10, pp. 1429-1433, Oct. 2022.T. Miyata, et al., “Random Undersampling Wireless EEG Measurement Device Using a Small TEG,” in Proc. IEEE Int. Symp. Circuits Syst. (ISCAS), May 2023. pp.1-5.T.-S. Chen, et al., “Low-complexity compressed-sensing-based watermark cryptosystem and circuits implementation for wireless sensor networks,” IEEE Trans. Very Large Scale Integr. (VLSI) Syst., vol. 27, no. 11, pp. 2485-2497, Nov. 2019.

[0006] However, in the method of Non-Patent Document 4, if the power of the digital watermark to be mixed is relatively large compared to the biometric signal, it may be possible to estimate what kind of digital watermark was used for masking by analyzing the measurement (interception) results for a number of times equal to or greater than the number L of digital watermark types. Conversely, if the power of the digital watermark is small compared to the biometric signal, it becomes close to the state where the biometric signal itself is being transmitted, which poses a problem that there is little point in mixing in a digital watermark.

[0007] The present invention has been made in consideration of the above, and aims to provide a sensing device, a data processing unit, and a sensing system that employ a noise masking cryptography method, which enables both low power consumption and robust security.

[0008] The sensing device of the present invention is a sensing device that periodically acquires a time-series target signal via a sensor and transmits it to the outside, and is equipped with a pseudo-noise generation unit that generates pseudo-noise based on seed information each time the target signal is periodically acquired, a pseudo-noise insertion unit that adds the generated pseudo-noise to the target signal, an electronic watermark generation unit that randomly selects a predetermined number of types of electronic watermarks each time the target signal is periodically acquired, an electronic watermark insertion unit that adds the selected electronic watermarks to the target signal, and a control unit that generates the pseudo-noise, and the control unit generates the seed information based on index information corresponding to the type of electronic watermark.

[0009] According to the present invention, a sensing device generates a seed value based on index information of a randomly selected digital watermark for a periodically acquired target signal, inserts pseudo-noise generated based on this seed value, inserts a randomly selected digital watermark, performs masking, encrypts, and transmits the signal to the outside. In this way, since it is sufficient to generate or select the pseudo-noise and digital watermark and add them to the target signal, encryption can be performed with a small amount of calculation. This enables low power consumption of the sensing device.

[0010] In addition, a data processing unit according to the present invention comprises a communication unit that receives a time-series target signal periodically transmitted from an external device, the target signal including pseudo-noise and a digital watermark; a pseudo-noise generation unit that generates pseudo-noise based on pre-associated seed information each time the target signal is periodically received, and a pseudo-noise removal unit that subtracts the generated pseudo-noise from the target signal; a correlation unit that, each time the target signal is periodically received, correlates the digital watermark contained in the target signal with a pre-prepared number of types of digital watermarks and extracts the digital watermark with the highest correlation; a digital watermark generation unit that generates the digital watermark extracted by the correlation unit each time the target signal is periodically received, and a digital watermark removal unit that subtracts the generated digital watermark from the target signal; and a control unit that controls processing in each of the above units, wherein the seed information is generated based on index information corresponding to the type of digital watermark.

[0011] According to this invention, each time a target signal is received, the target signal can have pseudo-noise corresponding to the seed information and an electronic watermark estimated through correlation processing removed, so that the signal can be decoded with relatively simple processing and a small amount of calculation.

[0012] The present invention also provides a sensing system comprising the sensing device and the data processing unit, and a noise masking cryptographic system using a randomly selected digital watermark and a pseudo-noise associated therewith.

[0013] According to the present invention, sensor data to be transmitted can be protected by a noise masking encryption method with a small amount of calculation, thereby making it possible to achieve both low power consumption and robust security.

[0014] FIG. 1 is an overall configuration diagram showing an embodiment of a sensing system according to the present invention. FIG. 1 is a diagram explaining the process of generating and adding pseudo-noise at the i-th and (i+1)-th times in a sensing device. FIG. 2 is a diagram showing an example of each signal applied to a sensing device. FIG. 3 is a diagram explaining the process of estimating and extracting the type of digital watermark inserted into a signal at the i-th time, and further generating pseudo-noise at the (i+1)-th time, in a data processing unit. FIG. 4 is a diagram explaining the first pseudo-noise and digital watermark insertion process for an electroencephalogram signal, performed on the sensing device side. FIG. 5 is a diagram explaining the first pseudo-noise and digital watermark subtraction process for a compressed signal, and the process of updating a seed value used in the second time, performed on the data processing unit side. FIG. 6 is a diagram explaining the second pseudo-noise and digital watermark insertion process for an electroencephalogram signal, performed on the sensing device side. FIG. 7 is a diagram explaining the second pseudo-noise and digital watermark subtraction process for a compressed signal, and the process of updating a seed value used in the third time, performed on the data processing unit side. FIG. 8 is a flowchart showing an example of the procedure of encryption processing executed in a sensing control unit of a sensing device. FIG. 9 is a flowchart showing an example of the procedure of decryption processing executed in a data processing control unit of a data processing unit. FIG. 10 is a diagram showing experimental results showing the relationship between the restoration accuracy, the magnitude of pseudo-noise, and the number of pairs when compressed sensing is used for compression and restoration.

[0015] 1 is a diagram showing the overall configuration of an embodiment of a sensing system 1 according to the present invention. The sensing system 1 includes a sensing device 10 that is attached to a living body, for example, to measure the state of a measurement target in a wearable manner, and a data processing unit 20 that incorporates a processor (CPU). First, the schematic configurations of the sensing device 10 and the data processing unit 20 will be described.

[0016] The sensing device 10 includes a sensor 11, a pseudo-noise inserter 12, a digital watermark inserter 13, a compressor 14, a communication unit 15, and a sensing controller 16. The sensor 11 can be adapted to various detection targets, and in this embodiment, it has electrodes for measuring biosignals, such as electroencephalogram (EEG) signals. The sensor 11 may be, for example, a pair of positive and negative electrodes attached to the human body, and periodically captures time-series EEG signals at a predetermined sampling rate, for example, 200 Hz, with each frame consisting of six seconds.

[0017] The pseudo-noise inserting section 12 generates pseudo-noise as will be described later each time a time-series electroencephalogram signal is periodically input, and inserts (adds) it into the electroencephalogram signal.

[0018] Each time an EEG signal with pseudo-noise inserted is input, the electronic watermark insertion unit 13 randomly selects one of multiple types of electronic watermark signals prepared in advance, as described below, using, for example, a random number generator, and inserts (adds) it into the biological signal.

[0019] The compression unit 14 performs compression processing (including random sampling assuming compressed sensing) on ​​the EEG signal into which the pseudo-noise and the digital watermark have been inserted, so as to obtain a data rate lower than the Nyquist frequency, for example. As will be described later, the generated compressed signal is restored, i.e., reconstructed, by a restoration unit 24 on the data processing unit 20 side.

[0020] Here, we will briefly explain compressed sensing. Compressed sensing involves multiplying a signal vector (e.g., an electroencephalogram signal) x with n elements by a measurement matrix Φ (m × n (m < n)) to obtain a compressed signal y (= Φx) with m elements while thinning it out. This reduces power consumption and enables longer-term sensing. However, under the condition of m < n, it is generally impossible to accurately reconstruct x from y and Φ (underdetermined system). However, when x can be expressed as the product of a dictionary matrix Ψ and a sparse vector s (i.e., containing many elements that are at level 0 or can be considered to be at level 0) (x = Ψs), there are various algorithms that can infer the sparse vector s from the compression matrix y and the sensing matrix θ (= ΦΨ), and it is known that the restored signal ^x can be obtained by multiplying the inferred vector ^s with the dictionary matrix Ψ (restoration algorithms: for example, BSBL (Block Sparse Bayesian Learning) and OMP (Orthogonal Matching Pursuit)).

[0021] The communication unit 15 may be wired, but here, for example, performs wireless communication over short distances, and Bluetooth, Wi-Fi, or the like can be applied to the data processing unit 20. Note that the communication unit 15 may be configured to be able to transmit and receive data to and from the communication unit 21 as needed. The sensing control unit 16 includes, for example, a processor (CPU), and reads and executes control programs stored in the storage unit 161 to control the sensing process of EEG signals, the generation process of pseudo-noise, the selection, generation process of digital watermarks, and the execution of compression processes in cooperation with each unit. The storage unit 161 stores the control programs required for the operation of each unit, the data required to execute the control programs, and the information required for compression.

[0022] The storage unit 161 also stores a program for generating a digital watermark shared with the data processing unit in advance, and a pseudo-noise time-series signal in accordance with parameters, for example, based on a Gaussian distribution. Index values ​​I corresponding to the types of digital watermarks are stored in the storage unit 161 as a correspondence table 1611 (see FIG. 2). The update process for the seed value S corresponding to the pseudo-noise is stored in the storage unit 161 as an update unit 1612 (see FIG. 2) indicated by an update formula.

[0023] The data processing unit 20 includes a communication section 21 capable of receiving a transmission signal from the communication section 15 , in this case a compressed signal, a pseudo-noise removal section 22 , a digital watermark removal section 23 , a restoration section 24 , and a data processing control section 25 .

[0024] Each time a compressed EEG signal is received, the pseudo-noise removal unit 22 generates pseudo-noise and subtracts it from the EEG signal, as described below. The digital watermark removal unit 23 estimates and selects one digital watermark and subtracts it from the EEG signal, as described below. The restoration unit 24 restores the compressed signal to its original signal based on the compressed sensing process described above. The data processing control unit 25 controls the processing in each unit, as described below.

[0025] 2 is a diagram illustrating the process of generating pseudo-noise for the i-th and (i+1)-th times in the sensing device 10. Note that Fig. 2 illustrates a state in which the i-th (present) electroencephalogram signal is input to the upper half, and the (i+1)-th (next) electroencephalogram signal is input to the lower half.

[0026] The pseudo-noise inserter 12 includes a pseudo-noise generator 121 and an adder 122. The pseudo-noise generator 121 generates an i-th seed value S ｉ The pseudo noise n corresponding to ｉ is generated and output to the adder 122. The digital watermark inserter 13 includes a digital watermark generator 131 and an adder 132. The digital watermark generator 131 generates a digital watermark w ｉ For example, the digital watermark wa is randomly selected as the pseudo-noise n ｉ and the EEG signal x with the digital watermark wa inserted. ｉis compressed by the compressor 14 to produce a compressed signal y ｉ is output to the communication unit 15 as

[0027] The digital watermark generation unit 131 randomly selects a digital watermark each time, while the pseudo-noise generation unit 121 generates a pseudo-noise n corresponding to a seed value S. The seed value S is set, for example, as follows. First, a digital watermark w is associated with an index value I. A correspondence table 1611 shows the correspondence between the digital watermark w and the index value I. A preset number of types of digital watermark w are prepared, and for example, index values ​​I are set as 1, 2, ..., 8 corresponding to eight types of digital watermarks wa, wb, ..., wL.

[0028] Then, the (i+1)th seed value S ｉ＋１ is the i-th seed value S ｉ The i-th index value I ｉ It is calculated by adding S ｉ＋１ = S ｉ +I ｉ The update unit 1612 of the (i+1)th pseudo-noise n ｉ＋１ is the seed value S ｉ＋１ The (i+1)th digital watermark w ｉ＋１ is selected randomly using a random number.

[0029] In this embodiment, power or root mean square (rms) is used to indicate the magnitude of a signal, pseudo noise, or digital watermark. For example, if there are n sampled signals for x1, i.e., x1[1], x1[2], ..., x1[n], the power P is P=(1 / N)*((x1[1]) ２ +(x1[2]) ２ +・・・+(x1[N]) ２ ) and the effective value is expressed as sqrt(P).

[0030] 3 is a diagram showing an example of various signals applied to the sensing device 10, showing the EEG signal x, pseudo-noise n, electronic watermark w, and their sum signal. In this example, the pseudo-noise n is set to approximately 200 μVrms so that masking is possible with a power greater than that of the EEG signal x. The electronic watermark w is set to approximately 60 μVrms, which is the same power as or slightly smaller than that of the EEG signal x. From the perspective of masking confidentiality, it is preferable that the pseudo-noise n be equal to or greater in power than the target signal.

[0031] 4 is a diagram illustrating the process of estimating and extracting the type of digital watermark inserted into the signal at the i-th time, and generating pseudo-noise at the (i+1)th time in the data processing unit. The pseudo-noise removal unit 22 includes a pseudo-noise generation unit 221 and a subtraction unit 222. The pseudo-noise generation unit 221 extracts the i-th seed value S ｉ The pseudo noise n corresponding to ｉ is generated and output to the subtraction unit 222. The digital watermark removal unit 23 includes at least a digital watermark generation unit 231 and a subtraction unit 232.

[0032] First, the process of the subtraction unit 222 will be described. The subtraction unit 222 subtracts the compressed signal y ｉ From the current seed value S ｉ The pseudo noise n corresponding to ｉ is subtracted to obtain the compressed signal y ｉ Then, the compressed signal y ｉ The digital watermark mixed in ' ｉ The data processing control unit 25 executes a process for removing the

[0033] The data processing control unit 25 includes a correlation unit 250. The correlation unit 250, in association with the digital watermark removal unit 23, calculates the compressed signal y ｉ , wL. The data processing control unit 25 calculates the correlation level, for example, a correlation coefficient, which indicates the degree of similarity between signals in each correlation process, and selects the watermark w that shows the highest correlation coefficient. ｉis estimated to be a digital watermark inserted on the sensing device 10 side and extracted. ｉ ' and the extracted digital watermark lol ｉ The subtraction process is performed by synchronizing the above two signals using a hardware synchronization processor 230 or software control.

[0034] In this embodiment, the sensing device 10 inserts pseudo-noise and a digital watermark into the raw EEG signal before compression, compresses it, and receives the compressed signal. Therefore, the data processing unit 20 uses a digital watermark compressed in the same manner as the compression unit 14 for correlation processing and subtraction processing.

[0035] The data processing control unit 25 has a storage unit 251, and stores an update unit 2512 similar to the update unit 1612 on the sensing device 10 side. That is, the update unit 2512 stores the next compressed signal y ｉ＋１ The pseudo noise n applied to ｉ＋１ More specifically, the watermark w that is estimated to have the highest correlation in the correlation process for the i-th signal is set. ｉ The index value I ｉ is extracted from the correspondence table 2511 in the storage unit 251, and then the current seed value S ｉ The (i+1)th seed value S ｉ＋１ Therefore, the pseudo-noise generator 221 sets the next pseudo-noise n ｉ＋１ is always updated with the seed value S ｉ＋１ Set based on.

[0036] Next, the signal processing of the sensing device 10 and the data processing unit 20 will be described with reference to FIGS. 5 to 8, with first and second examples.

[0037] 5 is a diagram illustrating the first pseudo-noise and digital watermark insertion process for the electroencephalogram signal, which is performed on the sensing device 10 side. １ is input, and the first pseudo noise n １is input to the adder 122. The waveform of the first pseudo-noise n, i.e., the initial seed value S, is set in advance and is shared between the sensing device 10 and the data processing unit 20. In FIG. 5, the seed value S １ = 1 is shown as an example of the initial value. １ can be random, and in FIG. 5, wb (index value I １ 5, the seed value S is selected by the digital watermark generating unit 131 and input to the adding unit 132. １ = 1, index value I １ = 2. The second seed value S ２ is S １ +I １ = 1 + 2 = 3. Then, the compression unit 14 compresses the compressed signal y １ is compressed and sent.

[0038] 6 is a diagram illustrating the first signal subtraction process for the compressed signal and the seed value update process used in the second cycle, which are performed on the data processing unit 20 side. １ is S １ = 1 in advance, the pseudo-noise n generated by the pseudo-noise generator 221 １ is output to the subtractor 222, and the compressed signal y １ '(=y １ -n １ ) is then output. １ ' is correlated with all the digital watermarks wa, wb, . . . wL in the correlation unit 250, and the digital watermark wb randomly set in the first round in FIG. 5 is estimated as the digital watermark with the highest correlation coefficient, and its index value I １ Therefore, in the first iteration, the update unit 2512 in FIG. 6 extracts the seed value S １ = 1, index value I １ = 2, so the second seed value S ２ is S １ +I １ =1+2=3. As mentioned above, this coincides with the contents of the update unit 1612 in FIG.

[0039] Next, Fig. 7 is a diagram illustrating the second signal insertion process for the electroencephalogram signal, which is performed on the sensing device 10 side. The sensing control unit 16 uses the second seed value S ２ The seed value S is generated by the pseudo noise generator 121 with reference to the pseudo noise value S = 3. ２ Pseudo noise n corresponding to =3 ２ to the adding unit 122. On the other hand, the sensing control unit 16 outputs the second digital watermark w ２ In the example of FIG. 7, the watermark wL (index value I ２ Therefore, in the second iteration, the seed value S ２ = 3, index value I ２ = 8. The third seed value S ３ is S ２ +I ２ =3+8=11.

[0040] Next, Fig. 8 is a diagram for explaining the second signal subtraction process for the compressed signal and the seed value update process used in the third cycle, which are performed on the data processing unit 20 side. First, as shown in Fig. 6, in the second cycle, the seed value S ２ = 3, so the seed value S ２ Pseudo noise n corresponding to =3 ２ is generated by the pseudo-noise generating unit 221 and output to the subtracting unit 222, and the electroencephalogram signal y ２ '(=y ２ -n ２ ) is output. Then, the electroencephalogram signal y ２ , wL is correlated with all the digital watermarks wa, wb, . . . wL in the correlation unit 250, and the digital watermark wL randomly set in the second round in FIG. 7 is estimated as the digital watermark w with the highest correlation coefficient, and its index value I ２ Therefore, in FIG. 8, the seed value S ２ = 3, index value I ２ = 8, so the third seed value S ３ is S ２ +I ２ =3+8=11. This is the third seed value S in FIG. ３matches.

[0041] Similarly, in each subsequent signal acquisition cycle, the sensing device 10 and the data processing unit 20 can always recognize the match between the pseudo-noise and the electronic watermark, and can perform the corresponding insertion and removal processes with high robustness.

[0042] 9 is a flowchart showing an example of the procedure of the encryption process executed by the sensing control unit 16 of the sensing device 10. First, it is determined whether it is the detection timing for the i-th period (i-th time) (step S1), and the process waits until the detection timing arrives. When the detection timing arrives, the digital watermark w ｉ is randomly selected (step S3). ｉ Synchronously with the seed value S ｉ The pseudo noise n corresponding to ｉ is generated and output to the adder 122, and then a randomly selected digital watermark w ｉ and outputs the result to the adder 132 (step S5). ｉ pseudo noise n ｉ and digital watermarks ｉ Encryption is performed by masking both noise signals.

[0043] Next, the update unit 1612 updates the seed value S ｉ And digital watermark lol ｉ The index value I i The seed value S for the (i+1)th period is ｉ＋１ (step S7) and return. ｉ and digital watermarks ｉ The detection signal xi into which the signal is inserted is compressed to produce a compressed signal y ｉ After generating the above, the process may return.

[0044] 10 is a flowchart showing an example of the procedure of the decoding process executed by the data processing control unit 25 of the data processing unit 20. First, the compressed signal y ｉ It is determined whether or not a signal is received (step S11), and the process waits until a signal is received. When a signal is received, the pseudo noise n corresponding to the seed value Si is generated. ｉ(in compressed form) and output to the subtraction unit 222 (step S13). ｉ Then, the signal y ｉ ' is subjected to correlation processing with all the digital watermarks wa to wL (in compressed format) (step S15), and the digital watermark w with the highest correlation coefficient is selected. ｉ (Step S17). The extracted digital watermark w ｉ The index value I ｉ The current seed value S ｉ The next seed value S ｉ＋１ (Step S19). The extracted digital watermark w ｉ (in compressed form) and generate the signal y ｉ ' and outputs it to the subtraction unit 232 (step S21). ｉ and digital watermarks ｉ is removed, and the encrypted electroencephalogram signal is decoded. The further compressed electroencephalogram signal is passed through the decompression unit 24 to be decompressed as a decompressed signal ^x ｉ will be restored to.

[0045] Figure 11 shows the experimental results showing the relationship between the restoration accuracy, the magnitude of the pseudo-noise, and the number of pairs p when compressive sensing is used for compression and restoration. The test data used was the FP1-F7 (frontal pole-anterior temporal) channel data of the CHB-MIT EEG signal test data, acquired with an electrode. The sampling frequency was 256 Hz, downsampled to 200 Hz, with one frame length set to 6 seconds. The size of the digital watermark was set to 60 μVrms.

[0046] In Figure 11, the horizontal axis represents the magnitude of the pseudo-noise (µVrms), and the vertical axis represents the normalized mean square error (NMSE). Each data point represents the NMSE of the restored signal relative to the original signal when received by this data processing unit (Receiver) and when restored using (1) to (4) for each number p of pairs received by an eavesdropper. Each data point represents the average for 1000 frames. The smaller the NMSE, the closer the signal is to the original signal and the more accurately it has been restored.

[0047] When received by this data processing unit (Receiver), the NMSE is constant at 0.11 regardless of the pseudo-noise level, indicating that decoding has been achieved with extremely high accuracy.On the other hand, in the case of an eavesdropper (1) to (4), although restoration is achieved to some extent in the region where the pseudo-noise power is small, the NMSE value increases as the pseudo-noise power increases, and it is clear that restoration becomes difficult when the pseudo-noise level is around 200 μVrms.

[0048] The present invention includes the following aspects: In this embodiment, the example of a biological signal has been described, but the present invention is not limited to this, and may include signals obtained by remotely sensing various types of environmental information, equipment status, the status of operating equipment, etc., or control signals for equipment, etc.

[0049] In this embodiment, power consumption is reduced by compressing the detection signal, pseudo-noise, and digital watermark all at once, but this may not be the case depending on the application. Furthermore, the data processing unit may remove the pseudo-noise and digital watermark after restoring the data. Since the pseudo-noise is removed before the digital watermark, the correlation of the digital watermark can be performed with higher accuracy.

[0050] Furthermore, in addition to the purpose of improving confidentiality, the present invention uses the digital watermark as an index for essentially randomly setting the next pseudo-noise via the index value, allowing the setting and removal of pseudo-noise to be performed with high robustness, but power savings can also be achieved by setting the power of the digital watermark to be approximately the same as that of the detection signal.

[0051] In addition, although the present embodiment shows an example in which one digital watermark is selected, the present invention is not limited to this. For example, a mode in which multiple digital watermarks are selected and used in combination may be adopted. Furthermore, a mode in which multiple pseudo-noises are similarly selected and used in combination may be adopted. This can improve confidentiality.

[0052] In addition, in the present embodiment, an example has been described in which signals are compressed on the sensing device side and decompressed on the data processing unit side, but depending on the application, regardless of whether it is wired or wireless, encryption and decryption only may be performed, and compression and decompression may not be necessary. In this aspect, it is possible to achieve more robust security for the sensing data to be transmitted by adopting a noise masking encryption method.

[0053] The update units 1612 and 2512 may also be configured to reset when the calculated seed value reaches a predetermined value. In this case, the set value can be suppressed to allow an appropriate number of pseudo-noise types to be expressed. Alternatively, the update units 1612 and 2512 may be configured to reset when the seed value has been updated a predetermined number of times, or at predetermined intervals, for example, on a daily basis. While a robust chain-based setting method is employed in which the next seed value is obtained by adding an index value to the previous seed value, other methods such as simple arithmetic operations may also be used. Furthermore, the seed value and index value do not have to be numerical values, and the next seed information may be determined based on the correlation between the two pieces of information.

[0054] Furthermore, in this embodiment, the addition units 122, 132, the subtraction units 222, 232, and the correlation unit 250 have been described as synchronizing the signals to be processed and processing them in time series, but this is not limited to this, and the signals to be processed may also be taken in once and addition, subtraction, and correlation calculations may be performed all at once.

[0055] Furthermore, the sensing device 10 may be configured with a pseudo-noise removal unit and a digital watermark removal unit, and the data processing unit 20 may be configured with a pseudo-noise insertion unit and a digital watermark insertion unit. Such a configuration makes it possible to achieve secure, low-power consumption bidirectional communication.

[0056] As described above, the sensing device of the present invention is a sensing device that periodically acquires a time-series target signal via a sensor and transmits it to the outside, and is equipped with a pseudo-noise generation unit that generates pseudo-noise based on seed information each time the target signal is periodically acquired, a pseudo-noise insertion unit that adds the generated pseudo-noise to the target signal, a digital watermark generation unit that randomly selects a predetermined number of types of digital watermarks each time the target signal is periodically acquired, a digital watermark insertion unit that adds the selected digital watermarks to the target signal, and a control unit that generates the pseudo-noise, and it is preferable that the control unit generates the seed information based on index information corresponding to the type of digital watermark.

[0057] According to the present invention, a sensing device generates a seed value based on index information of a randomly selected digital watermark for a periodically acquired target signal, inserts pseudo-noise generated based on this seed value, inserts a randomly selected digital watermark, performs masking, encrypts, and transmits the signal to the outside. In this way, since it is sufficient to generate or select the pseudo-noise and digital watermark and add them to the target signal, encryption can be performed with a small amount of calculation. This enables low power consumption of the sensing device.

[0058] Furthermore, the control unit preferably includes an updating unit that updates the seed information for the next cycle by associating the index information corresponding to the type of digital watermark randomly selected in the current cycle with the index information corresponding to the type of digital watermark randomly selected in the current cycle. With this configuration, the control unit can apply a so-called chain mechanism, in which the seed information for the next cycle is updated by associating the index information corresponding to the type of digital watermark randomly selected in the current cycle with the seed information corresponding to the pseudo-noise generated in the current cycle. In this way, the security level of the target signal can be improved with a small amount of calculation between the seed information and the index information, enabling both low power consumption and robust security. Furthermore, applying the so-called chain mechanism to generate the pseudo-noise eliminates the need to directly transmit a decryption key to the receiving side, thereby eliminating accidents associated with key transmission and enabling more stable decryption.

[0059] In addition, it is preferable that the index information and the seed information are numerical values, and the update unit adds the seed value of the current cycle to the index value of the current cycle to obtain the seed value of the next cycle. With this configuration, the calculation load in encryption and decryption can be light.

[0060] Preferably, the update unit resets the seed value when the seed value reaches a predetermined value. With this configuration, an upper limit is set for the value of the seed information, and the number of types of pseudo-noise that can be generated can be set to a required number.

[0061] Furthermore, it is preferable that the power of the pseudo-noise is at least equal to or greater than the power of the target signal. With this configuration, even if the signal power of the digital watermark is kept low, it is possible to mask the target signal with the pseudo-noise, thereby improving encryption performance.

[0062] Furthermore, the present invention preferably includes a compression unit that compresses the pseudo-noise, the digital watermark, and the target signal all at once to a low data rate. With this configuration, the compression operation can be performed in one go, thereby achieving power saving.

[0063] Furthermore, the data processing unit of the present invention comprises a communication unit that receives a time-series target signal periodically transmitted from an external device, the target signal including pseudo-noise and a digital watermark; a pseudo-noise generation unit that generates pseudo-noise based on pre-associated seed information each time the target signal is periodically received, and a pseudo-noise removal unit that subtracts the generated pseudo-noise from the target signal; a correlation unit that, each time the target signal is periodically received, correlates the digital watermark contained in the target signal with a pre-prepared number of types of digital watermarks and extracts the digital watermark with the highest correlation; a digital watermark generation unit that generates the digital watermark extracted by the correlation unit each time the target signal is periodically received, and a digital watermark removal unit that subtracts the generated digital watermark from the target signal; and a control unit that controls the processing in each of the above units, wherein it is preferable that the seed information is generated based on index information corresponding to the type of digital watermark.

[0064] According to this invention, each time a target signal is received, the target signal can have pseudo-noise corresponding to the seed information and an electronic watermark estimated through correlation processing removed, so that the signal can be decoded with relatively simple processing and a small amount of calculation.

[0065] Furthermore, it is preferable that the control unit includes an updating unit that updates the seed information for the next cycle by associating the index information corresponding to the type of the digital watermark extracted by the correlation unit in the current cycle. According to this configuration, the seed information for the next cycle is updated by estimating the digital watermark that is the most highly correlated among all the digital watermarks in the current cycle with the current seed information, and associating the index information corresponding to the estimated digital watermark with the current seed information. Therefore, in the next cycle, pseudo-noise corresponding to the updated seed information is generated.

[0066] In the present invention, it is also preferable that the pseudo-noise removal unit and the digital watermark removal unit are connected in this order to a stage subsequent to the communication unit. With this configuration, correlation of the digital watermark is calculated after the pseudo-noise is removed, thereby improving correlation accuracy.

[0067] Furthermore, a sensing system according to the present invention preferably includes the sensing device and the data processing unit. According to this invention, a noise masking cryptosystem can be constructed using a randomly selected digital watermark and a pseudo-noise associated therewith.

[0068] REFERENCE SIGNS LIST 1 sensing system 10 sensing device 12 pseudo-noise insertion unit 121, 221 pseudo-noise generation unit 122, 132 addition unit 13 digital watermark insertion unit 131, 231 digital watermark generation unit 14 compression unit 16 sensing control unit 161, 251 storage unit 1611, 2511 correspondence table 1612, 2512 update unit 20 data processing unit 21 communication unit 22 pseudo-noise removal unit 222, 232 subtraction unit 23 digital watermark removal unit 25 data processing control unit 250 correlation unit

Claims

1. In a sensing device that periodically acquires a time-series target signal via a sensor and transmits it externally, the sensing device includes a pseudo-noise generation unit that generates pseudo-noise based on seed information every time the target signal is periodically acquired, and a pseudo-noise insertion unit that adds the generated pseudo-noise to the target signal, an electronic watermark generation unit that randomly selects and generates a preset number of types of electronic watermarks every time the target signal is periodically acquired, and an electronic watermark insertion unit that adds the generated electronic watermark to the target signal, and a control unit that selects the pseudo-noise, wherein the control unit generates the seed information based on index information corresponding to the type of the electronic watermark.

2. The sensing device according to claim 1, wherein the control unit includes an update unit that updates the seed information for the next cycle by associating the index information corresponding to the type of the electronic watermark randomly selected in the current cycle.

3. The sensing device according to claim 2, wherein the index information and the seed information are numerical values, and the update unit adds the seed value in the current cycle and the index value in the current cycle to obtain the seed value for the next cycle.

4. The sensing device according to claim 3, wherein the update unit resets the seed value when the seed value reaches a predetermined numerical value.

5. The sensing device according to claim 1, wherein the power of the pseudo-noise is at least equal to or greater than the power of the target signal.

6. The sensing device according to any one of claims 1 to 5, further including a compression unit that compresses the pseudo-noise, the electronic watermark, and the target signal together at a low data rate.

7. A communication unit that receives a target signal in a time series periodically transmitted from the outside, the target signal including pseudo-noise and a digital watermark; a pseudo-noise generation unit that generates pseudo-noise based on pre-associated seed information every time the target signal is received periodically, and a pseudo-noise removal unit that subtracts the generated pseudo-noise from the target signal; a correlation unit that, every time the target signal is received periodically, calculates the correlation between the digital watermark included in the target signal and a number of types of digital watermarks prepared in advance, and extracts the digital watermark with the highest correlation; a digital watermark removal unit that, every time the target signal is received periodically, has a digital watermark generation unit that generates the digital watermark extracted by the correlation unit, and subtracts the generated digital watermark from the target signal; and a control unit that controls the processing in each unit, wherein the seed information is generated based on index information corresponding to the type of digital watermark, the data processing unit.

8. The data processing unit according to claim 7, wherein the control unit includes an update unit that updates the seed information for the next cycle by associating the index information corresponding to the type of digital watermark extracted by the correlation unit in the current cycle.

9. The data processing unit according to claim 7, wherein the pseudo-noise removal unit and the digital watermark removal unit are connected in this order at the subsequent stage of the communication unit.

10. A sensing system including the sensing device according to claim 1 and the data processing unit according to claim 7.

Images