Underwater communication system, transmitter and receiver

The SSMS method in underwater communication systems addresses demodulation errors by dividing transmission data into bits and using waveform patterns for cross-correlation, ensuring stable communication despite Doppler shifts and multipath interference without complex calculations.

JP7593575B2Active Publication Date: 2024-12-03MITSUBISHI ELECTRIC DEFENSE & SPACE TECHNOLOGIES CORP +1
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
JP2023024828
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2023-02-21
Publication Date
2024-12-03
Estimated Expiration
2043-02-21

AI Technical Summary

Technical Problem

Existing underwater communication systems face challenges with demodulation errors due to Doppler shift and multipath interference, which affect frame synchronization and accurate Doppler shift estimation, leading to time position shifts and increased computational load.

Method used

The system employs a method called Stretchable Symbols with Multiple Sequences (SSMS) where transmission data is divided into bits, and waveform patterns are generated based on the value and order of each divided data, allowing for cross-correlation function calculations without requiring Doppler shift estimation and correction, thus stabilizing demodulation.

Benefits of technology

This approach enables stable demodulation without Doppler shift estimation and correction, reducing computational load while maintaining accurate communication even in environments with varying Doppler shifts and multipath interference.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007593575000001
    Figure 0007593575000001
  • Figure 0007593575000002
    Figure 0007593575000002
  • Figure 0007593575000003
    Figure 0007593575000003
Patent Text Reader

Abstract

To provide an underwater communication system for performing stable communication without performing Doppler shift estimation and correction on a reception signal.SOLUTION: A transmitter for performing underwater communication generates a transmission signal 5a which is a digital signal corresponding to transmission data to be transmitted, where the digital signal is configured such that from any of 00, 01, 10, 11 shown by values of division data obtained by dividing transmission data, which is a digital data to be transmitted, for every two bits and transmission order of the division data, waveform patterns determined for the division data are arranged in the transmission order, and a transmission interval between the waveform patterns is made to be Tint. The transmitter converts the generated transmission signal 5a to an analog signal to generate a transmission wave signal and transmits the generated transmission wave signal toward a receiver.SELECTED DRAWING: Figure 5
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present disclosure relates to an underwater communication system, a transmitter and a receiver for use in underwater communications. The present disclosure relates to synchronization of transmitted data and received data in underwater communications, and countermeasures against Doppler shift in underwater communications. [Background technology]

[0002] Conventional techniques include Patent Documents 1 to 4, which relate to synchronization of transmitted data and received data, or techniques for dealing with Doppler shift in underwater communications. Patent Document 1 shows a method of improving the deterioration of correlation characteristics by sequentially selecting a spreading code from among multiple spreading codes for each symbol using modulation with spreading codes and providing a dummy section at the boundary of the sequentially selected spreading codes. In reception, frame synchronization is performed only once with the first symbol of a communication frame, and the timing of extraction of subsequent symbols from the received signal is fixed symbol length.

[0003] Patent Document 2 shows a method of calculating an autocorrelation function from a synchronization symbol sequence located at the beginning of a communication frame and obtaining frame synchronization from the peak value of the autocorrelation function. Frame synchronization is performed only once at the beginning of the communication frame.

[0004] In Patent Document 3, the amount of Doppler shift is estimated from a reference signal in a communication packet, and Doppler shift estimation and correction are performed by a method of applying correction to the estimated frequency of an information signal.

[0005] In Patent Document 4, the amount of Doppler shift is estimated using a pilot signal inserted in an OFDM signal, and Doppler shift correction is performed in two stages: resampling conversion and frequency deviation correction. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2011-151436 A [Patent Document 2] JP 2006-109279 A [Patent Document 3] JP 2020-68500 A [Patent Document 4] JP 2017-195566 A Summary of the Invention [Problem to be solved by the invention]

[0007] In Patent Document 1, frame synchronization is performed only once at the first symbol of a communication frame during reception, and the timing of extraction of subsequent symbols from the received signal is fixed. Therefore, when the received signal expands or contracts in time due to Doppler shift, a shift occurs between the desired reception timing. Therefore, the time position shift cannot be ignored at the rear of the communication frame, and demodulation errors occur.

[0008] In Patent Document 2, frame synchronization is performed only once at the beginning of a communication frame. Therefore, when the received signal expands or contracts in time due to Doppler shift, a shift occurs between the signal delimiter position for the succeeding QPSK symbol sequence in the communication frame and the desired reception timing. Therefore, the time position shift cannot be ignored at the rear of the communication frame, and demodulation errors occur.

[0009] Regarding Patent Document 3, in an environment with strong sound wave reflection, underwater acoustics suffers from multipath interference, and the amount of Doppler shift may differ between direct waves and reflected waves. When a received signal is affected by both the Doppler shift and multipath interference, accurate Doppler shift estimation becomes difficult.

[0010] The process in Patent Document 4 is complicated and requires a large amount of calculation, which causes a problem of a large load on the receiver. [Means for solving the problem]

[0011] The underwater communication system according to the present disclosure is an underwater communication system including a transmitter and a receiver. The transmitter includes: a transmission signal generating unit that generates a transmission signal, the transmission signal being a digital signal corresponding to the transmission data, the transmission signal being a digital signal in which the transmission data is divided into data pieces of the same number of bits, the waveform patterns being determined for each of the divided data pieces from their values ​​and transmission orders, and the transmission signals being arranged in the transmission order, and transmission intervals being provided between the waveform patterns; a wave transmitter that transmits the transmission signal converted into an analog signal as a wave transmission signal that is an acoustic wave; Equipped with The receiver includes: A receiver for receiving the transmitted signal; The transmitted signal is converted into a digital signal. Contains multiple symbols From the received signal, For each symbol, the symbol waveform is adjusted so that the symbol length is located in the center of the extraction section. The extracted signal is Cut a signal extraction unit for extracting the a correlation calculation unit which regards each of all bit data obtained from the same number of bits as a value of each divided data, regards the number of times the cut-out signal is cut out as the transmission order, uses a plurality of the waveform patterns determined from the value of each divided data and the transmission order as reference signals, and performs cross-correlation function calculation between the cut-out signal and each reference signal; a correlation comparator that determines, from a result of the cross-correlation function calculation, a reference signal corresponding to the cut-out signal among the reference signals, and determines a bit value indicated by the cut-out signal and a reception time of the cut-out signal based on the determined reference signal; Equipped with. Effect of the Invention

[0012] According to the underwater communication system according to the present disclosure, a received signal in which a Doppler shift has occurred can be demodulated without demodulation errors, without requiring calculation-intensive Doppler shift estimation and correction. [Brief description of the drawings]

[0013] [Figure 1] FIG. 1 is a diagram of a first embodiment, showing a configuration of an underwater communication system 3. [Diagram 2]FIG. 1 is a diagram showing a block configuration of a transmitter 1 according to a first embodiment. [Diagram 3] FIG. 2 is a diagram showing a block configuration of a receiver 2 according to the first embodiment. [Figure 4] FIG. 1 is a diagram of the first embodiment, which diagrammatically illustrates a received signal y(k) of a communication frame in which S consecutive symbols are included. [Diagram 5] FIG. 2 is a diagram of the first embodiment, showing a transmission signal 5a having a communication frame configuration in which the SSMS system is adopted. [Figure 6] FIG. 2 is a diagram of the first embodiment, showing an overview of modulation processing in the transmitter 1. [Figure 7] FIG. 1 is a diagram of the first embodiment, and is a flowchart showing an overview of a modulation process in a transmitter 1 and a flowchart showing an overview of a demodulation process in a receiver 2. [Figure 8] FIG. 2 is a diagram of the first embodiment, showing an overview of demodulation processing in a receiver 2, focusing on a received signal. [Figure 9] FIG. 11 is a diagram according to the first embodiment, showing a correlation calculation method for an extracted signal corresponding to symbol number 1. [Figure 10] FIG. 10 is a diagram of the first embodiment, showing received codes and time positions of extracted signals corresponding to symbol numbers. [Figure 11] FIG. 11 is a diagram of the first embodiment, showing correlation function calculation for an extracted signal corresponding to symbol number 1. [Figure 12] FIG. 12 is a diagram of the first embodiment, showing a correlation function calculation in which sound wave reflection is added to the communication simulation of FIG. 11 . [Figure 13] FIG. 1 is a diagram of the first embodiment, showing a method for identifying the location of a transmitter 1 using multiple receivers 2. [Figure 14] FIG. 10 is a diagram of the first embodiment, showing a waveform pattern in which the sampling rate is converted according to the amount of Doppler shift. [Figure 15] FIG. 11 is a diagram of a second embodiment, showing a hardware configuration in which a transmitter 1 and a receiver 2 are implemented using a microcontroller 110. [Figure 16]FIG. 11 is a diagram of the second embodiment, showing a program configuration when the transmitter 1 and the receiver 2 are implemented using a microcontroller 110. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] In the description of the embodiments and the drawings, the same elements and corresponding elements are given the same reference numerals. Descriptions of elements given the same reference numerals are omitted or simplified as appropriate. In the following embodiments, "unit" may be read as "circuit", "step", "procedure", "process" or "circuitry" as appropriate.

[0015] Embodiment 1 ***Configuration Description*** FIG. 1 shows the configuration of an underwater communication system 3. The underwater communication system 3 communicates underwater using sound waves. The underwater communication system 3 uses ultrasonic waves as the sound waves, but the sound waves are not limited to ultrasonic waves. The underwater communication system 3 includes a transmitter 1 and a receiver 2. The transmitter 1 transmits ultrasonic waves as a transmission signal, which is a digital signal converted into an analog signal. The receiver 2 receives the transmission signal underwater using the receiver 8 and demodulates it.

[0016] FIG. 2 shows a block diagram of the transmitter 1. The transmitter 1 includes a data division unit 4, a transmission waveform generation unit 5, a DA conversion and amplification unit 6, and a transmitter 7. The data division unit 4 is a division unit. The transmission waveform generation unit 5 is a transmission signal generation unit. The data division unit 4 divides the transmission data, which is data consisting of 0 and 1 such as "0111001001110010", into pieces of the same number of bits. If the number of identical bits is 2, the data division unit 4 divides "0111001001110010" into pieces of 2 bits such as 01, 11, 00, 10, 01, 11, 00, 10. The number of identical bits can be the number of bits indicated by the symbol unit. The number of bits indicated by the symbol unit is the amount of information contained in one symbol. For example, assuming QPSK, the four codes contained in one symbol are (00)(01)(10)(11). The symbol unit of QPSK is modulated into four phase amounts of 2 bits: (00), (01), (10), and (11). In the following explanation, the number of bits used by the data division unit 4 to divide the transmission data is the information amount of the symbol unit, and is 2 bits.

[0017] The transmission waveform generating section 5 generates a transmission signal 5a. The generation of the transmission signal 5a will be described later.

[0018] The DA conversion and amplification unit 6 converts the transmission signal 5a into an analog signal to generate and amplify a transmission wave signal 6a. The wave transmitter 7 transmits the transmission wave signal 6a.

[0019] FIG. 3 shows a block diagram of the receiver 2. The receiver 2 includes a receiver 8, an amplifier / AD converter 9, a reference waveform generator 10, multiple correlation calculators 11, a correlation comparator 12, a signal extractor 13, and a data combiner 14. The operation of each component will be described later. The reference waveform generator 10 is a reference signal generator. The amplifier / AD converter 9 converts the transmitted signal received as a received signal into a digital received signal. The multiple correlation calculators 11 may be written as correlation calculators 11-1, 11-2, 11-3, and 11-4 when necessary to distinguish them. Although the correlation calculators 11-1, 11-2, 11-3, and 11-4 are shown individually in FIG. 3, the functions of the correlation calculators 11-1, 11-2, 11-3, and 11-4 may be provided by a single correlation calculator 11.

[0020] In an underwater communication system 3 that uses ultrasonic signals as a communication means, a Doppler shift occurs when a transmitter 1 and a receiver 2 are moving at high speed. The change in sound wave frequency due to the Doppler shift is expressed by Equation 1. f'=f(1+△) (Formula 1) The △ in Equation 1 indicates the amount of Doppler shift, which is the ratio of the speed of sound to the relative movement speed. The amount of Doppler shift is expressed as the expansion and contraction of a signal in the time domain. Regarding the expansion and contraction of a signal, the received signal y(t') is expressed by Equation 2 with respect to the transmitted signal x(t'). y(t')=x((1+△)t') (Equation 2) The sampling period T of transmitter 1 and receiver 2 s The received signal that has been Doppler shifted with respect to is expressed by Equation 3. y(nTs)=x((1+△)nTs) (Equation 3) The time series of a discrete signal is expressed as t=nTs.

[0021] FIG. 4 shows a schematic representation of a received signal y(t) when a communication frame has a configuration having S consecutive symbols. As shown in FIG. 3, the received signal is the output signal of the amplification and AD conversion unit 9. Here, a set of four codes (00), (01), (10), and (11) expressed in 2 bits is called a symbol. When there is no Doppler shift, that is, when △=0, the receiver 2 performs frame synchronization at the beginning of the communication frame. In this case, the frame synchronization position by the receiver 2 is set to t=0. The transmitter 1 and receiver 2 communicate with each other using a symbol length T sym Receiver 2 knows the symbol length T sym The time position of the last symbol in the communication frame is t = (S-1) × T sym That is, the last signal division is t=(S-1)×T sym This is the position at .

[0022] In Fig. 4, when the received signal is stretched due to the Doppler shift (△<0) and when the received signal is compressed (△>0), the apparent symbol length is T' sym , and T'' sym The symbol length T sym If the receiver 2 performs signal division in the same way as in the case of no Doppler shift △=0 based on t=(S-1)×T, a shift occurs between the signal division position and the actual symbol position. This shift becomes noticeable at the rear of the communication frame. As shown in Figure 4, sym If receiver 2 extracts the signal with symbol number S at the time position, the result will be as follows: A signal with a different symbol number will be extracted, as in the middle row △<0, or the signal will be extracted outside the communication frame range, as in the bottom row △>0, and demodulation will fail.

[0023] Patent technology 1 and patent technology 2 only perform frame synchronization at the beginning of the communication frame. Therefore, when there is a Doppler shift, demodulation failure may occur due to a time position shift, as shown in the middle row of Figure 4 (△<0) or the lower row (△>0). Patent technology literature 3 and patent technology literature 4 show a method for estimating and correcting the Doppler shift. However, in estimating the Doppler shift, it is necessary to take into consideration that the Doppler shift amount of the direct wave and the reflected wave differs due to multipath interference. Due to multipath interference, the receiver receives a direct wave and a combination of many reflected waves. The reflected wave has a different relative speed from the direct wave due to reflection on the sea surface, seabed, and obstacles, so the Doppler shift amount is also different. Accurate Doppler shift estimation is difficult under multipath interference. Also, sampling rate conversion is used to correct the Doppler shift. However, sampling rate conversion requires a large amount of calculation for linear interpolation processing that changes the time position for each sample of the discrete signal and resampling processing using a digital filter, which increases the load on the receiver.

[0024] The underwater communication system 3 of the first embodiment deals with the Doppler shift using a method that does not depend on Doppler shift estimation and correction. The method of the underwater communication system 3 makes it possible to perform appropriate demodulation processing even when a Doppler shift occurs.

[0025] FIG. 5 shows a communication frame structure in which a method called SSMS (Stretchable Symbols with Multiple Sequences) is applied, which is adopted in the first embodiment, in the underwater communication system 3. The symbol lengths of the symbols indicated by symbol numbers 1 to S are "T sym The SSMS communication frame shown in Figure 5 has a symbol interval T int This is a frame structure with a space between the first and second frames.

[0026] ***Explanation of Operation*** (SSMS method explanation) The operation of generating the transmission signal 6a using the SSMS method by the transmitter 1 will be described. The configuration of the transmitter 1 of the underwater communication system 3 is as shown in FIG. 2. It is assumed that "0111001001110010" is input as transmission data to the data division unit 4. The data division unit 4 divides the input transmission data by the amount of information per symbol. That is, while it is assumed that the amount of information per symbol is 2 bits, the data division unit 4 divides "0111001001110010" into eight pieces: 01, 11, 00, 10, 01, 11, 00, and 10. Each of these eight pieces of data is called divided data. Each of these divided data is a transmission code.

[0027] Although 2 bits are used as an example, if the divided data in symbol units is N bits, the transmission code, which is the divided data, is M=2 N In this case, M indicates the number of coding modulation levels.

[0028] (Transmission waveform generator 5) The transmission waveform generating unit 5 generates a transmission signal 5a. The transmission signal 5a is a digital signal in which the transmission data, which is a digital signal to be transmitted, is divided into data of the same number of bits, and waveform patterns determined for each divided data based on the values ​​and transmission order of each divided data are arranged in transmission order, with transmission intervals provided between the waveform patterns. The transmission signal 5a is a digital signal corresponding to the transmission data. A specific description will be given below. The transmission waveform generating unit 5 stores in advance in a storage device a number of waveform patterns equal to "MS pieces x number of sample data H" for the modulation multi-level number M, the number of symbols S in a communication frame, and the number of sample data H in one symbol as a waveform pattern group 160. Note that in reality, there are MS composite waves in which waveforms of the number of sample data H (waves) are added together. The number of symbols S is the total number of divided data in a communication frame. The number of sample data H is determined by the "symbol length T sym / sampling period T s The waveform pattern group 160 generated by the transmission waveform generating section 5 will be described in detail below with reference to FIG.

[0029] FIG. 6 is a diagram showing an outline of the modulation process in the transmitter 1. As shown in FIG. The left side of FIG. 7 is a flowchart showing an outline of the modulation process in the transmitter 1.

[0030] (Step S11) In step S11, the transmission waveform generating unit 5 generates in advance a waveform pattern group 160 in which sine waves obtained by phase-modulating a plurality of carrier waves are synthesized. For example, the number of carrier waves is set to the number of sample data H in one symbol length. In FIG. 6, the number of sample data H is set to 2000 samples, and the transmission waveform generating unit 5 prepares 2000 sine waves having frequencies that are equal divisions of the sample frequency. Specifically, in the example of FIG. 6, the number of sample data H in one symbol length T sym = 10 ms, sampling frequency Fs = 200 kHz, sampling period T s = 5 × 10^-6 ms. Therefore, the number of sample data in one symbol is H = 2000 (waves). The frequency interval between each carrier is Fs / H, which is 100 Hz.

[0031] (Step S12) In step S12, a phase shift is performed in the range of -π to π according to the pseudorandom numerical value on each of the 2000 carrier waves, carrier wave 1 to carrier wave 2000. As shown in Fig. 6, the pseudorandom numerical values ​​ρ(s, c) (c = 1 to 2000) applied to carrier waves 1 to 2000 at a certain symbol number s are different values.

[0032] (Step S13) In step S13, the transmission waveform generating unit 5 first generates a waveform pattern for the symbol unit (00) of symbol number 1 (first symbol) as follows: The transmission waveform generating unit 5 generates a composite wave by combining (adding) 2000 waves of phase-shifted carrier wave 1 to carrier wave 2000. The transmission waveform generating unit 5 sets this composite wave as a waveform pattern of one symbol length for the symbol unit (00) of the first symbol.

[0033] (Step S14) In step S14, the transmission waveform generating unit 5 similarly generates waveform patterns for [number of coded multi-level modulations (M) × number of symbols (S)] by changing the pseudo-random numerical value ρ. That is, as shown in the waveform pattern group 160 in Fig. 6, the transmission waveform generating unit 5 similarly generates waveform patterns from the symbol unit (01) of the first symbol to the symbol unit (11) of the final symbol.

[0034] (Step S15) In step S15, the transmission waveform generating unit 5 selects a waveform pattern corresponding to 2-bit data extracted by the data dividing unit 4 from the bit string of the transmission data from the waveform pattern group 160. The transmission waveform generating unit 5 selects a symbol interval T int are inserted and concatenated to generate the transmission signal 5a. Specifically, as shown in the lower right of FIG. 6, when the transmission data is divided into 2 bits, and is "01, 11, 00, 10, 01, 11, 00", the transmission waveform generating unit 5 selects a corresponding waveform pattern from the waveform pattern group 160 according to the rank division data [s:m] described later. The transmission waveform generating unit 5 selects a waveform pattern determined by the symbol number s and the division data m from the waveform pattern group 160. In FIG. 6, the symbol number s is the value of s in the s-th symbol of the first symbol, the second symbol, to the final symbol (S-th symbol), and the division data m is (00), (01), (10), and (11). Since the coding modulation multi-level number M=4, the division data m, i.e., the transmission code 4a, is one of (00), (01), (10), and (11). When the transmission data is "01, 11, 00, 10, 01, 11, 00", the waveform pattern (first symbol, 01) is selected at the beginning (01), the waveform pattern (second symbol, 11) is selected at the next (11), and so on. The transmission waveform generator 5 arranges the selected waveform patterns in order of the number of symbols s, and also selects the symbol interval T int is inserted and the whole is concatenated to generate one frame of transmission waveform.

[0035] (Symbol interval T int (Details of Symbol Interval T int In the no-signal section corresponding to the symbol interval T, I samples of data in which all the I sample data are zero are inserted, and a transmission signal 5a for one communication frame is generated. int / sampling period T s "

[0036] (Rank split data) The ranked split data, consisting of a symbol number (transmission order of the split data) and a transmission code (split data), is as follows. In the example, the eight split data pieces, 01, 11, 00, 10, 01, 11, 00, 10, are transmitted in the order 01, 11, 00, 10, 01, 11, 00, 10 from the left. The symbol number is the rank number in the transmission order of the split data. Alternatively, it is also a number that indicates the order of division. Specifically, for the symbol numbers, 01 is number 1, 11 is number 2, 00 is number 3, 10 is number 4, and so on, with the last divided data 10 being number 8. If the rank divided data is expressed as [symbol number s: divided data m], the transmission waveform generator 5 generates the following eight: [1:(01)], [2:(11)], [3:(00)], [4:(10)], [5:(01)]...[8:(10)].

[0037] The transmission waveform generating unit 5 generates and stores in advance MS waveform pattern groups 160 corresponding to all combinations [s:m] of symbol number s (s=1, 2, 3, 4, ..., 8) and divided data m (m=(00), (01), (10), (11)). This waveform pattern group 160 is shown in FIG. 6. If the first divided data is (10), the transmission waveform generating unit 5 selects from the waveform pattern group 160 a waveform pattern determined by [1:(10)], and if the second divided data is (11), it selects from the waveform pattern group 160 a waveform pattern determined by [2:(11)].

[0038] The transmission signal 5a finally generated by the transmission waveform generating unit 5 has the communication frame configuration shown in Figure 5. The box with symbol number 1 in Figure 5 is a waveform pattern corresponding to the rank division data of [1:(ij)]. Also, the black box with symbol number S on the right end is a waveform pattern corresponding to the rank division data of the final transmission symbol [S:(ij)] in the communication frame. Here, i and j are either 0 or 1. The S waveform patterns to be concatenated are all different waveform patterns. This is self-evident, since, as described in steps S11 and S12, the selected MS waveform patterns are all generated based on different pseudo-random numbers, and further, the order of selection from the waveform pattern group is also selected in symbol order.

[0039] Each waveform pattern of the waveform pattern group 160 shown in FIG. 6 may be H samples of data filtered by a band pass filter (hereinafter, BPF) in order to suppress unnecessary waves other than the frequency band required by the transmitter 1. Specifically, in the case of a communication simulation described later, the frequency band of the sound waves used is set to 10 kHz to 30 kHz in consideration of the characteristics of the transmitter 7, receiver 8, and underwater acoustic propagation path, and a waveform pattern is generated by a composite wave in which carrier waves of unnecessary frequency bands other than this frequency band are cut. Alternatively, although not shown, a BPF may be physically added in the transmitter 1 in order to suppress unnecessary waves other than the frequency band used as described above, and the signal may be filtered by the BPF before being used as the transmission signal 5a. The transmission signal 5a shown in FIG. 5 is digital-to-analog converted in the DA conversion and amplification unit 6 and amplified by a transmission amplifier, and then transmitted as a transmission signal 6a.

[0040] The operation of the receiver 2 will be described. The configuration of the receiver 2 of the underwater communication system 3 is as shown in FIG. 1. The receiver 8 receives the transmission signal 6a transmitted from the transmitter 7 as a reception signal. The amplifier / AD converter 9 amplifies and converts the reception signal into an analog / digital signal, and converts it into a digital reception signal 9a. The amplifier / AD converter 9 samples the reception signal 9a at 200 kHz, the same as the sampling period of the transmission waveform data. In order to suppress unnecessary waves other than the frequency band required for the receiver 2, the signal may be filtered by a BPF (not shown) before being used as a reception signal. The correlation calculator 11 calculates the cross-correlation function between the extracted signal waveform extracted from the reception signal and a reference signal consisting of M waveform patterns. Here, M is the number of coded multilevels of the symbol, and if N bits are the amount of information of one symbol, M is expressed as M=2 N It becomes.

[0041] FIG. 8 shows an overview of the demodulation process in the receiver 2, focusing on the received signal. 7 is a flowchart showing an outline of the demodulation process in the receiver 2. The upper part of FIG. 8 shows the process by the signal extractor 13, the middle part shows the calculation process by the correlation calculator 11, and the lower part shows the calculation results by the correlation calculator 11.

[0042] (Step S21) In step S21, the amplifier / AD converter 9 detects the sampling period T s The received signal 9a is sampled at the same 200 kHz.

[0043] (Step S22) As shown in the upper part of FIG. 8, background noise, multipath interference waves, and Doppler shift are superimposed on the received signal 9a compared to the transmitted signal 5a. In step S22, a BPF (not shown) is a filter having a pass band of the aforementioned operating frequency of 10 to 30 kHz, and removes unnecessary waves outside that band.

[0044] (Step S23) The signal extracting unit 13 extracts a signal as an extracted signal from the received signal 9a obtained by converting the transmitted wave signal 6a into a digital signal. In S23, as shown in the upper part of Fig. 8, the signal extraction unit 13 ideally extracts a waveform of one symbol from the received signal 9a as an extraction signal so that each one-symbol length of the received signal 9a is located at the median value as the time position of the extraction section. That is, there are cases where the time position of the received signal shifts forward or backward for each symbol, such as when the received signal expands or contracts due to Doppler shift, or when the signal level of the direct wave is too low to be received and only delayed waves due to multipath are observed. To deal with these cases, the signal extraction unit 13 inserts a symbol interval T on both sides of the waveform of one symbol so that the waveform of one symbol is always present within the extraction section. int Extract the signal so that

[0045] (Step S24) In S24, the signal extractor 13 estimates the extraction position for the first symbol by any one of the following methods (1) to (3), and determines the extraction section of the first symbol. (1) When the transmitter and receiver are asynchronous, an analog watchdog circuit that converts sound pressure into a voltage is provided, that is, a circuit that constantly detects a certain signal level or above and detects the input of a received signal, and the output of this circuit is used as a trigger to determine the cut-out position. (2) Alternatively, the time position of the first symbol is determined by sequentially sliding the extraction section, including the signal-free period before the reception of the first symbol, and constantly correlating with the reference data of the first symbol to detect peaks. (3) Alternatively, by synchronizing the time between transmitters 1 and 2, the transmission time and the underwater propagation path delay time are obtained from the position information of both transmitters (based on the UUV's IMU information), and the optimal position for transmitting the first symbol is determined. For the second symbol and onwards, the extraction position is determined from the correlation peak position for the immediately preceding symbol.

[0046] (Step S25) The correlation calculation unit 11 regards all bit data obtained from the same number of bits as the value of each divided data, and regards the number of times the cut-out signal is cut out as the transmission order. The correlation calculation unit 11 uses multiple waveform patterns determined from the assumed value of each divided data and the assumed transmission order as reference signals, and performs cross-correlation function calculation between the cut-out signal and the waveform patterns that are each reference signal. The correlation calculation unit 11 uses, as the reference waveform pattern, a waveform data group that is the same as the waveform pattern group 160 stored in the transmitter 1. To give a specific example, the reference waveform generation unit 10 has the waveform pattern group 160 shown in the lower left of Fig. 6 as the reference waveform pattern. In step S25, as shown in the middle of Fig. 8, the correlation calculation unit 11 generates cross-correlation function calculation values ​​between the cut-out signal and four types of reference waveform patterns. The correlation calculation unit 11 multiplies the cut-out signal by the reference waveform pattern while shifting the reference waveform pattern by one sample data, and calculates the sum. This is the state shown in the middle of Fig. 8.

[0047] (Step S26) The correlation comparator 12 determines the reference signal corresponding to the cut-out signal from among the reference signals based on the cross-correlation function calculation results of each correlation calculator, and determines the bit value indicated by the cut-out signal and the reception time of the cut-out signal based on the determined reference signal. Specifically, the correlation comparator 12 selects the one with the largest PAPR (Peak to Average Power Ratio) from among the four cross-correlation function calculation values ​​corresponding to the symbol codes (00), (01), (10), and (11) to determine the symbol code. For example, when the cross-correlation function calculation value of the reference waveform pattern corresponding to the symbol code 01 is selected, the symbol code is determined to be 01. The lower part of FIG. 8 shows the one with the most prominent peak among the four cross-correlation function calculation values. The correlation comparator 12 uses the correlation peak position to determine the next cut-out section.

[0048] FIG. 9 shows a method of signal extraction by the signal extraction unit 13 and a method of calculating a cross-correlation function for rank division data [1:(ij)] of symbol number 1. The signal extracted from the received signal 9a by the signal extraction unit 13 is called an extracted signal. It is assumed that the time position of the reference signal pulse in the time reference signal can be considered to be the same in the transmitter 1 and the receiver 2. The time reference signal is generated by time synchronizing clocks (not shown) built into the transmitter 1 and the receiver 2 of the opposing underwater communication system 3 with which it communicates, using a 1PPS (Pulse Per Second) signal that can be obtained from a GPS receiver, for example. The transmission signal 5a is transmitted as a communication frame in accordance with the time reference signal. The reception signal 9a is delayed by the arrival time of the transmission signal 6a that arrives at the receiver 8 from the transmitter 7. The extraction section of symbol number 1 for the reception signal 9a is a period from the time reference signal generation point at which the transmission signal of symbol number 1 is transmitted as a communication frame to the maximum arrival time of the sound wave + T sym +T int Interval T containing c1 In the section T c1includes the time of the maximum arrival time of the sound wave, which is the time of "relative distance between transmitter 1 and receiver 2 / speed of sound in water". Since the speed of sound is an almost fixed value, this maximum arrival time depends on the relative distance. If one or both of transmitter 1 and receiver 2 are mounted on a moving object such as a UUV, the maximum relative distance is calculated from the maximum moving range and the section T including the maximum arrival time is c1 In this way, the section T c1 is set in advance in the signal extraction unit 13. The accumulated error over time between the clocks of the transmitter 1 and the receiver 2 appears as a pulse time position shift of the time reference signal and a position shift from the ideal position of the extraction section for the received signal 9a, which will be described later. However, in the underwater communication system 3, as shown in FIG. 5, the transmitter 1 sym The waveform pattern has a symbol interval T int A communication frame with a symbol interval T before and after the waveform pattern is generated as the transmission signal 5a. int However, it absorbs the time position shift caused by the accumulated error between clocks over time. In addition, there are cases where the time position of the received signal shifts forward or backward for each symbol, such as when the received signal is stretched by Doppler shift, or when the signal level of the direct wave is too low to be received and only the delayed wave due to multipath is observed. In these cases, the symbol interval T int However, this absorbs the time position shift. Therefore, it is possible to include all of the received signals corresponding to the symbol numbers within the extracted section of the received signal 9a.

[0049] (reference signal) For the received signal corresponding to symbol number 1, a cross-correlation function calculation between M reference signals is performed by a plurality of correlation calculation units 11. In FIG. 9 showing an example, the number of reference signals corresponds to the number of coded modulation levels, and M=2. 2= 4. The number of symbols to be transmitted is 8 (S = 8). The reference waveform generating unit 10 stores, as a reference signal, a waveform pattern group identical to the waveform pattern group 160 (FIG. 6) possessed by the transmitter 1. Here, the reference signal is a waveform pattern selected from the waveform pattern group 160 by the rank division data [s:m]. Note that s in the rank division data [s:m] is the same as the number of cut-outs k in the cut-out signal. Therefore, the receiver knows the value of s, that is, the value of k. However, m is unknown. Therefore, the known k is converted into k * Then, receiver 2 will be [k * :(00)],[k * :(01)],[k * :(10)],[k * :(11)] is selected from the waveform pattern group and used as the reference signal. Specifically, it is as follows. The reference signal for the kth extracted signal is a waveform pattern corresponding to each of [k:(00)], [k:(01)], [k:(10)], and [k:(11)]. Here, in the case of k=1, this corresponds to the first symbol in the waveform pattern group 160 at the lower left of FIG. 6, so the four waveform patterns (00), (01), (10), and (11) in FIG. 6 become the reference signal. This is because the kth extracted signal is a signal that includes a waveform pattern corresponding to the kth transmission order and any of the rank division data of bits (00), (01), (10), and (11). The reference waveform generating unit 10 provides waveform patterns corresponding to rank division data [k:(00)], [k:(01)], [k:(10)], and [k:(11)] corresponding to k in the kth extraction to each correlation calculating unit 11. If k=1, the reference waveform generating unit 10 provides four types of waveform patterns, from (00) to (11), of the first symbol from among the waveform pattern group 160 shown in the lower left of FIG. 6, to each correlation calculating unit 11.

[0050] (Cross-correlation function calculation) In the receiver 2 of FIG. 3, if the number of cutouts is k=1, the following process is performed. The reference waveform generating unit 10 obtains the following as the rank division data: To correlation calculation unit 11-1 [1:(00)], To correlation calculation unit 11-2 [1:(01)], To correlation calculation unit 11-3 [1:(10)], [1:(11)] is input to correlation calculation unit 11-4. Also, the reference waveform generating unit 10 inputs the waveform pattern of (00) of the first symbol at the bottom left of FIG. 6, which corresponds to the rank division data [1:(00)], to the correlation calculation unit 11-1 as a reference signal. Similarly, the reference waveform generating unit 10 inputs the waveform pattern of (01) of the first symbol at the bottom left of FIG. 6, which corresponds to the rank division data [1:(01)], to the correlation calculation unit 11-2 as a reference signal. Similarly, the reference waveform generating unit 10 inputs the waveform pattern of (10) of the first symbol at the bottom left of FIG. 6, which corresponds to the rank division data [1:(10)], to the correlation calculation unit 11-3 as a reference signal. Similarly, the reference waveform generating unit 10 inputs the waveform pattern of (11) of the first symbol at the bottom left of FIG. 6, which corresponds to the rank division data [1:(11)], to the correlation calculation unit 11-4 as a reference signal. Each of correlation calculation units 11-1 to 11-4 performs a cross-correlation function calculation between the cut-out signal corresponding to symbol number 1 and the input reference signal, and outputs the cross-correlation function as the result of the calculation. Taking a method using a convolution operation on the time axis as an example, the cross-correlation function calculation in correlation calculation units 11-1 to 11-4 is performed as follows: A zero-extended waveform pattern is generated for the reference signal by adding zero data after the waveform pattern that is the reference signal so that the number of data items is the same as the number of sample data items in the cut-out section cut out from the received signal.

[0051] 8, the correlation calculation unit 11 multiplies the waveform pattern, which is the reference signal, by the sample data string (extracted signal) in the extracted section for each sample element to obtain a sum. The correlation calculation unit 11 sets the absolute value as the first data of the cross-correlation function calculation value. Next, the correlation calculation section 11 shifts the entire sample data string (reference signal) of the zero-extended waveform pattern backward by one sample, and embeds zero data in the leading sample data. This is used as the zero-extended waveform pattern of the new reference signal, and the correlation calculation unit 11 multiplies the sample data sequence by the sample data sequence of the cut-out section for each sample element, calculates the sum, and uses the absolute value as the second data for the cross-correlation function calculation value.

[0052] 8, the correlation calculation unit 11 also calculates the third and subsequent cross-correlation function calculation values ​​by shifting the entire zero-extended waveform pattern of the reference signal and inserting zero data to generate a new zero-extended waveform pattern of the reference signal and calculate the cross-correlation function values. In this way, each correlation calculation unit 11 outputs, as a data string, the same number of cross-correlation function calculation values ​​as the number of sample data in the cut-out section. As shown in the lower part of Fig. 8, a large correlation function peak is naturally observed for the cut-out signal of the received signal corresponding to the transmission code at the time of transmission and the reference signal corresponding to that transmission code, since they originate from the same waveform pattern. In other words, the correlation function peak appears when the waveform pattern of the cut-out signal from the received signal corresponding to the rank division data matches that of the reference signal.

[0053] The correlation comparator 12 compares the cross-correlation function peaks calculated by the correlation calculators 11-1 to 11-4. The cross-correlation function peaks can be detected using the PAPR (Peak to Average Power Ratio) method, which uses the average value of the data string of the cross-correlation function values ​​as a reference to suppress comparison errors. That is, the ratio of the peak level to the average value of the data string of the cross-correlation function values ​​is set as the PAPR value. The correlation comparator 12 identifies the received code with the highest PAPR value from the comparison result of the PAPR values, and outputs the received code to the data combiner 14. The data combiner 14 performs demodulation processing to combine multiple received codes and output the received signal 9a as received data 14a.

[0054] In addition, the time position indicated by the cross-correlation function peak when the correlation comparator 12 identifies the received code of the first extracted signal indicates the sound wave arrival time. Therefore, the correlation comparator 12 converts the sound wave arrival time τ1 into the distance between the transmitter 1 and the receiver 2, and outputs the converted distance.

[0055] In Fig. 9, the received code and time position for the cut-out signal of symbol number 1 are as follows. The received code is code 1 (01 in bit representation), and the time position is τ1. In other words, the first cut-out signal was [1:(01)]. Note that the cross-correlation function calculation in correlation calculation units 11-1 to 11-4 can employ a method using convolution calculation on the time axis as described above, or a method in which the signals are converted to the frequency domain by FFT, multiplied, and then converted back to the time domain by IFFT.

[0056] (Extraction from the second time onwards) 10 shows a method for calculating the cross-correlation function of received symbol divided data from symbol number 2 onward. Using the time position τ1 of symbol number 1 obtained from the cross-correlation function calculation for symbol number 1 as a reference, signal extraction unit 13 extracts symbol divided data from received signal 9a.

[0057] As shown in FIG. 5, in the transmission signal 5a, the symbol length is T sym , the symbol interval is T int As shown in FIG. 10, the section cut out by the signal cutout unit 13 is based on the reception time position of symbol number 1, T sym <t≦2(T sym +T int ) Therefore, Cut length = T sym +2T int It is. It is an absolute requirement that the extraction section be set so that the received signal of symbol number 2 is always included. For this reason, the received signal of symbol number 2 is set as close to the center position of the extraction section as possible. In other words, even when the received signal of symbol number 2 is expanded by the Doppler shift, the trailing T int By deleting the symbol number 2, the received signal is always included. int Even if the T is shortened by the Doppler shift, intis removed, the received signal with symbol number 2 is always included. When a Doppler shift occurs in this way, as shown in Figure 4, the signal is stretched or compressed, causing a time position shift of the symbols. Furthermore, even when the signal level of the direct wave is too low to be received and only the multipath delayed wave is observed, the multipath generally arrives with a delay, and even when this multipath delayed wave is used as the received signal and is subject to correlation calculations, a time position shift occurs. Furthermore, a time position shift occurs due to the accumulated error over time between the time reference signal of transmitter 1 and the time reference signal of receiver 2. Since the SSMS method uses symbol spacing, the time position shift of the symbols is T int The following is acceptable:

[0058] The received signals of the symbols before and after the cut-out section of symbol number 2 are (1) to (3) as follows. (1) Delayed wave due to multipath of symbol number 1. (2) The signal component of the received signal of symbol number 1 that has been stretched by the Doppler shift. (3) The signal component of symbol number 2 and the T int This is the signal component at the beginning of the received signal with symbol number 3, which appears when the Ideally, these unnecessary signal components should be set so as not to enter the cut-out section. The SMSS method uses correlation processing to eliminate the effects of these unnecessary waves, making the presence of these unnecessary waves acceptable.

[0059] After the signal extraction unit 13 extracts the received signal for symbol number 2, the process is the same as for symbol number 1. That is, the reference waveform generating unit 10 obtains the following as rank division data: To correlation calculation unit 11-1 [2:(00)], To correlation calculation unit 11-2 [2:(01)], To correlation calculation unit 11-3 [2:(10)], [2:(11)] is input to correlation calculation unit 11-4. At the same time, the reference waveform generating unit 10 inputs the waveform pattern of the second symbol (00) at the bottom left of FIG. 6, which corresponds to the rank division data [2:(00)], to the correlation calculation unit 11-1 as a reference signal. The reference waveform generating unit 10 inputs the waveform pattern of the second symbol (01) at the bottom left of FIG. 6, which corresponds to the rank division data [2:(01)], to the correlation calculation unit 11-2 as a reference signal. Similarly, the reference waveform generating unit 10 inputs the waveform pattern of the second symbol (10) at the bottom left of FIG. 6, which corresponds to the rank division data [2:(10)], to the correlation calculation unit 11-3 as a reference signal. Similarly, the reference waveform generating unit 10 inputs the waveform pattern of the second symbol (11) at the bottom left of FIG. 6, which corresponds to the rank division data [2:(11)], to the correlation calculation unit 11-4 as a reference signal. Correlation calculation units 11-1 to 11-4 calculate the cross-correlation function between the cut-out signal for symbol number 2 and each reference signal. At this time, it should be noted that the waveform patterns of the reference signals for symbol number 1 and symbol number 2 are all different. In other words, for the reference signals for symbol number 1 and symbol number 2, the waveform patterns corresponding to the eight signals [1:(00)], [1:(01)], [1:(10)], [1:(11)], [2:(00)], [2:(01)], [2:(10)], and [2:(11)] are all different.

[0060] In an environment with strong acoustic reflection, the waveform of symbol number 1 may interfere with the direct wave of symbol number 2 as a delayed wave due to multipath interference. When the waveform patterns of each symbol number are different, signal components that do not correspond to the received code can be suppressed as uncorrelated components when calculating the cross-correlation function. Therefore, the SSMS method of the first embodiment is resistant to multipath interference and ambient noise (noise).

[0061] In FIG. 10, the received code and time position for symbol number 2 are as follows. The received code for symbol number 2 is code 3 (11 in bit representation), and the time position is τ2. The correlation comparator 12 identifies code 3 and time position τ2 as the received code, and outputs code 3 to the data combiner 14. The data combiner 14 concatenates bit representation 11 corresponding to code 3 to the end of bit representation 01 of symbol number 1 as a demodulation process, and configures it as a received frame. From symbol number 3 onwards, a process similar to the process for the extracted signal corresponding to symbol number 2 is repeated as a demodulation process for the extracted signal of symbol number 3. That is, a signal is extracted using the time position calculated by the cross-correlation function calculation in the demodulation process of the previous symbol number, and a cross-correlation function calculation is performed on the extracted signal.

[0062] As described above, in the SSMS system, the receiver 2 performs demodulation processing while determining the time position for each symbol (each rank-divided data). Therefore, there is no need for Doppler shift estimation and correction processing for the received signal. In addition, the SSMS system can also handle non-stationary fluctuations in which the Doppler shift amount △ switches between positive and negative within a communication frame.

[0063] Fig. 11 is a calculation example of the correlation function for the extracted signal of symbol number 1 in a communication simulation. In the communication simulation, the sound speed was 1,500 m / s, the sampling frequency was 200 kHz, the frequency band was 10 kHz to 30 kHz, the symbol length was 10 ms, the symbol interval was 10 ms, the amount of information per symbol was 2 bits, the modulation multi-level number was 4, the number of symbols in a communication frame was 100, and the communication frame length was 2 seconds. The sound wave propagation conditions were a relative movement speed of 2.57 m / s (9.36 km / h) when the Doppler shift occurred, and an input S / N ratio of 10 dB. The correlation function φ for code 2 1,2 Since (τ) shows the highest correlation peak value, received code 2 and a distance of 100 m (arrival time 66.5 ms) are output.

[0064] Figure 12 shows an example of the calculation of the correlation function when sound wave reflection is added to the communication simulation in Figure 11. The sound wave reflectance of the sea surface and the seabed is set to 1.0 and 0.7, respectively, and the correlation function is calculated under the condition that multipath interference in which direct waves and delayed waves are superimposed is further added. The correlation function φ of code 2 due to multipath interference 1,2 Only (τ) shows multiple peaks. The multiple peaks are caused by the correlation function φ 1,2 Since only (τ) is used, the received signal can be correctly demodulated even in situations where the received signal is distorted by Doppler shift and multipath interference.

[0065] The transmission waveform generating unit 5 of the transmitter 1 determines the symbol interval T int The transmission waveform generator 5 may adjust the length of the symbol interval T int It is also effective to make T variable instead of fixed. For example, when the relative movement (△<0) occurs in the direction of separation, the amount of Doppler shift increases, and the reception signal 9a appears to be extended, causing the reception waveform period to become "T sym +T int ", the transmission waveform generator 5 uses the symbol interval T int Symbol interval T based on int In addition, in an underwater ultrasonic propagation environment in which the delay and strength of multipath waves are large and overlap with the next symbol reception waveform, the transmission waveform generating unit 5 makes the symbol interval T int Symbol interval T based on int In this case, the transmission rate is sacrificed. However, the symbol interval T int When the next symbol reception waveform cannot be correctly demodulated when is fixed, the overlap with the unnecessary wave disappears and demodulation becomes possible. Conversely, when the Doppler shift is small, or when there is relative movement (△>0) in the approaching direction, and the received signal appears to be compressed, or when the delay and strength of the multipath wave are small, the symbol interval T int It is possible to reduce the number of bits and improve the transmission rate.

[0066] In addition, when the signal-to-noise ratio (SNR) in the underwater ultrasonic propagation environment is small, the symbol length T sym Measures can be taken to increase the number of samples by lengthening the time or by raising the sampling frequency. These measures can increase the SNR per symbol and reduce demodulation errors. As in the past, the modulation multilevel number M is selected according to the underwater ultrasonic propagation environment, and the frequency bandwidth is selected to an optimal value according to the hardware characteristics of the DA conversion and amplification unit 6, the transmitter 7, the receiver 8, and the amplification and AD conversion unit 9.

[0067] (Positioning) Regarding conventional underwater acoustic positioning technology, for example, the following technology is described in the Transactions of the Institute of Electronics, Information and Communication Engineers, Vol. A, J65-A A3, pp. 221-228. (1) That is, ultrasonic signals from an underwater sound source are received by multiple receivers. (2) The direction of incidence (direction cosine) of the ultrasonic signal is estimated from the phase difference between the receivers. (3) The distance between the sound source and the receiver (slant range) is estimated from the propagation time of the sound wave, and the position of the sound source relative to the receiver is estimated using these two. These positioning methods (1) to (3) are called SSBL (Super Short Baseline).

[0068] A case where the SSBL method is applied to the underwater communication system 3 of the first embodiment will be described. Fig. 13 is a diagram showing a method for determining the position of a transmitter 1 using multiple receivers 2. The two receivers in Fig. 13 show a method for determining the position of a transmitter 1 using a receiver group 16 consisting of multiple receivers 2. For example, if the distance between two receivers 8-1 and 8-2 is L, the angle of arrival is θ, the time difference of arrival is δT, and the speed of sound is C, the relationship is expressed by Equation 4, that is, the angle of arrival θ can be calculated as Equation 5. The two receivers 8-1 and 8-2 are assumed to be located on the same horizontal plane, i.e., at the same water depth. δT×C=L×cosθ (Formula 4) θ=cos -1 (δT×C÷L) (Formula 5) The distance L between the receiver 8-1 and the receiver 8-2 constituting the receiver group 16 is a known fixed value. The two receivers 2-1 and 2-2 are arranged in the vicinity of each other as the receiver group 16 so as to synchronize the sampling timing. This allows ΔT to be measured with the resolution of the sampling period, and the arrival time difference ΔT can be obtained with high accuracy. As a result, the arrival angle θ can be obtained with high accuracy. In addition, the position of the transmitter 1 based on the receiver group 16 can be calculated as the distances ra and rb output by the correlation comparator 12 of the receivers 2-1 and 2-2 of the receiver group 16, and the position on the polar coordinate of the arrival angle θ. The underwater communication system 3 outputs the distances ra and rb at short intervals in symbol units. Therefore, high-frequency positioning is possible, and accuracy can be improved by statistical methods. The underwater communication system 3 equipped with the receiver group 16 composed of multiple receivers 2 has the characteristic that positioning can be performed simultaneously with data communication.

[0069] For ease of explanation, the two-receiver group 16 in Fig. 13 only describes azimuth estimation in the horizontal plane. However, the three receivers in Fig. 13 show the following. Receivers 8-3 and 8-4 are placed on the horizontal plane, and receiver 8-5 is placed vertically. With this arrangement, the direction in the depth direction can be calculated by the same calculation process as for two receivers (horizontal plane). In this way, if the position of the receiver group 16 is known, the two-dimensional or three-dimensional position of the transmitter 1 can be estimated.

[0070] In the explanation of Fig. 13, a method of estimating position by preparing multiple receivers 2 in the underwater communication system 3 has been described. However, it is also possible to realize an implementation method in which only the receivers 8 and the amplifier / AD conversion units 9 of the receiver 2 are prepared multiple times, and the other parts of the receiver 2 are implemented so that a series of received signal processes are time-division processed, thereby sharing hardware such as computing units and reducing the hardware scale.

[0071] (Effect of relative movement speed) The reference waveform generating unit 10 performs a stretching / shrinking transformation to expand or shrink the waveforms of a plurality of reference signals. The reference waveform generating unit 10 is a reference signal adjusting unit. The correlation calculating unit 11 performs a cross-correlation function calculation using each of the stretched / shrinked reference signals. This will be specifically described below. 11, the relative moving speed between the transmitter 1 and the receiver 2 is relatively slow, so there is no need to convert the sampling rate of the reference signal. If the relative moving speed is even faster, the reference waveform generating unit 10 additionally stores a waveform pattern whose sampling rate has been converted in addition to the stored waveform pattern group. In this way, a plurality of waveform patterns corresponding to a plurality of Doppler shift amounts are prepared as reference signals, and are stored in the memory of the reference waveform generating unit 10 together with the reference signal, which is the original waveform pattern. That is, in the example, the reference waveform generating unit 10 generates waveform patterns corresponding to [s:(00)], [s:(01)], [s:(10)], and [s:(11)] for s=1 to S in advance as waveform patterns, and stores them in an auxiliary storage device or a main storage device. Here, if [s:(00)], [s:(01)], [s:(10)], and [s:(11)] for s=1 to S are one group G, a group of waveform patterns is prepared for each Doppler shift amount for this one group. That is, a group G#i is prepared for a Doppler shift amount #i. The group G#i is obtained by performing sampling rate conversion on each waveform pattern of the group G. The cross-correlation function is calculated for each of these reference signals in a round-robin manner using the method described above, and demodulation is performed. This makes it possible to demodulate even when the relative moving speed is even greater.

[0072] FIG. 14 shows a method of preparing a waveform pattern that has been stretched and converted in advance by changing the sampling rate according to the amount of Doppler shift by resampling processing in the waveform pattern of the reference waveform generating unit 10. Let Γa and Γb be waveforms that correspond to the symbol lengths before and after the Doppler shift, respectively. When the amount of Doppler shift becomes large as a negative value (△<0), that is, when the received signal is stretched, the stretched waveform pattern Γa<Γb is used as the reference waveform. Conversely, when the amount of Doppler shift becomes positive (△>0) and the received signal is shrunk, a waveform pattern Γa>Γb in which the shrunk portion is padded with zeros is prepared. In this way, when generating multiple reference signals, the reference waveform generating unit 10 performs stretching and contracting conversion to stretch or contract the waveform of the reference signal that is used as a reference. The correlation calculation unit 11 calculates the cross-correlation function between the cut-out signal and each reference signal that has been stretched and converted.

[0073] The memory area for the reference signal of the waveform pattern (△=0) is MS × (symbol length T sym / sampling period T s) was required. However, the required memory area increases according to the assumed Doppler shift amount. For example, the relative movement speed is positive for approaching movement and negative for separating movement. Assume that reference signals are prepared to cover the range of ±Vknot in 1 knot increments, i.e., -Vknot···-10knot, -9knot,···-1knot, 0knot, 1knot,···9knot, 10knot···Vknot. In this case, memory for sample data of "(2V+1)MS×H samples" is required. Furthermore, in the case of waveform pattern Γa<Γb where △<0, memory area for the signal extension, that is, {H samples of sample data} / {△ / (1+△)} is required. However, these reference signals are prepared in advance in the memory used by reference waveform generating unit 10 of receiver 2. As a result, receiver 2 does not need to perform sampling rate conversion of received signals in real time, and the calculation load during operation can be reduced. When a peak appears in the cross-correlation function calculation value with the waveform pattern that has been stretched and converted at the closest relative moving speed among the relative moving speeds prepared for ±V knots in 1 knot increments, the corresponding code is selected. This makes it possible to correctly demodulate even when a large Doppler shift occurs due to a large relative moving speed.

[0074] We also explain the case where the relative moving speed can be estimated by receiver 2 using a DVL (Doppler velocity log), an inertial unit (IMU) and a GPS on a moving platform such as a UUV (Unmanned Undersea Vehicle) or a ship equipped with a transmitter 1 and a receiver 2. Assume that the absolute positions, ground speeds, and moving directions of the transmitter 1 and receiver 2 are successively measured. The transmitter 1 periodically transmits these pieces of measurement information as communication data from the transmitter 1 to the receiver 2. The receiver 2 can then successively calculate the relative movement speed. In this case, the receiver 2 can use the calculated relative moving speed to select only the waveform pattern created with the closest or several nearby relative moving speeds from among the waveform patterns prepared for ±V knots in 1 knot increments, and use them as a reference signal. For example, if the calculated relative moving speed is around 5 knots, the reference waveform generating unit 10 of the receiver 2 selects only 5 knots, or only waveform patterns around 4 knots, 5 knots, and 6 knots, and sends them to each correlation calculation unit 11. Each correlation calculation unit 11 performs cross-correlation function calculation. In this way, it is no longer necessary to perform cross-correlation function calculations in a brute-force manner, and it is possible to reduce the number of cross-correlation function calculations. Even when the relative moving speed is large, that is, when the Doppler shift amount is large, it is possible to suppress the communication error rate and maintain the positioning performance without significantly increasing the amount of cross-correlation function calculations.

[0075] Furthermore, in the above-mentioned method of "transmitter 1 periodically transmitting measurement information", a case will be described in which absolute position information is periodically transmitted as communication data from transmitter 1 to receiver 2. In this case, it will be described that the amount of calculation can be reduced by having receiver 2 sequentially calculate the relative distance, or by using the relative distance obtained in the immediately preceding cross-correlation function calculation process, etc.

[0076] Maximum arrival time of sound wave, including L / C, T c1 Unknown big T c1 Instead of using it as a reference, T according to the relative distance L' calculated sequentially in the current situation is used. c1 Using this, it is possible to narrow the cut-out section of the received signal of the first symbol. C is the speed of sound. As a result, in the first communication frame immediately after communication is started, T c1 needs to be provided to the signal extraction unit 13 from the outside in advance, but from the second communication frame onwards, it is possible to narrow the extraction section using the relative distance L' obtained immediately before. This makes it possible to reduce the amount of calculation.

[0077] ***Advantages of the First Embodiment*** (1) In the underwater communication system 3, even if the received waveform expands or contracts due to the Doppler shift, the signal extraction unit 13 extracts the corresponding received waveform for each symbol (each rank-divided data) at the most appropriate signal extraction position. Then, the correlation calculation unit 11 executes correlation calculation. Then, the correlation comparison unit 12 determines a likely extraction position of the received waveform corresponding to the next symbol from the correlation peak position. Therefore, the underwater communication system 3 can perform stable acoustic communication without performing Doppler shift estimation and correction, which require a large amount of calculation. (2) In addition, even when the amount of Doppler shift fluctuates significantly within a communication frame or when the amount of Doppler shift differs between the direct wave and the reflected wave due to multipath interference, the effects are reduced and demodulation can be performed without being affected by the Doppler shift. (3) Also, the symbol interval T int By adjusting the length of the antenna, the effects of Doppler shift, which occurs depending on the relative moving speed between the transmitter 1 and the receiver 2, and Doppler shift and multipath interference, which occur in the underwater ultrasonic propagation environment, can be suppressed by trading off the information transmission rate. (4) In the first embodiment, a waveform pattern is prepared by expanding or contracting the waveform of the transmission signal according to a plurality of assumed Doppler shift amounts, and the waveform pattern is used as a reference waveform. Even when there is a large relative moving speed between the transmitter 1 and the receiver 2, the received signal can be demodulated without being affected by the Doppler shift. (5) Furthermore, the underwater communication system 3 functions as a positioning system by including multiple receivers 2 with receivers 8 arranged at different positions. In other words, by simultaneously and frequently performing multiple distance measurements for each symbol unit in the communication frame, it becomes possible to perform highly accurate positioning of the two-dimensional or three-dimensional position of the transmitter 1 relative to the receiver 2.

[0078] Embodiment 2 As a second embodiment, a hardware configuration and a program configuration in the case where the transmitter 1 and the receiver 2 are implemented using a microcontroller 110 will be described with reference to Fig. 15 and Fig. 16. Fig. 15 shows the hardware configuration, and Fig. 16 shows the program configuration.

[0079] 15, a commercially available microcontroller, for example, a microcontroller with the model name STM32L4R5AI from STMicroelectronics, can be used. The microcontroller 110 is a device that executes a transmission program 131 and a reception program 231. The microcontroller 110 includes a CPU core 111 and an internal memory 120. The microcontroller 110 is also a System On Chip IC (Integrated Circuit) that integrates units called internal peripherals, such as a DMA controller 130, an RTC 140, a DA converter 610, and an AD converter 930. DMA stands for "Direct Memory Access" and RTC stands for "Real Time Clock."

[0080] Inside the microcontroller 110, the CPU core 111 and the DMA controller 130 access various resources via an internal bus (not shown). The CPU core 111 and the DMA controller 130 access the built-in flash memory 150 of the built-in memory 120, the built-in work SRAM 170, the DA converter 610, the AD converter 930, the RTC 140, and a control register (not shown) in the DMA controller 130.

[0081] (CPU core 111) The CPU core 111 performs various calculations according to successively called machine language instructions written in the transmission program 131 or the reception program 231 in the built-in flash memory 150. When performing various calculations, the CPU core 111 accesses various resources and performs various calculations using a fixed point arithmetic unit and an FPU (Floating Point Unit) in the CPU core 111.

[0082] (Memory) The microcontroller 110 includes an internal flash memory 150 and an internal working SRAM 170 as the internal memory 120. The internal flash memory 150 stores the transmission program 131 or the reception program 231, and the waveform pattern group 160. The capacity of the internal flash memory 150 is assumed to be 2 MByte. If the program size or the waveform pattern group 160 increases and cannot be accommodated in 2 MByte, an external flash memory may be used separately.

[0083] (Wave pattern group 160) Regarding the waveform pattern group 160, the number of bits per sample data is set to 12 bits to match the 12-bit input width of the built-in flash memory 150 and the DA converter 610. The waveform pattern group 160 expressed in binary format as amplitude values ​​is stored in the built-in flash memory 150. In the case of the communication simulation case of the first embodiment, the waveform pattern group 160 is 2000 samples×100 symbols×4 multi-level modulation number×12 bits=1.2 MByte. Of the 2 MByte of the built-in flash memory 150, 0.8 MByte excluding the 1.2 MByte of the waveform pattern group 160 is used to store the transmission program 131 or the reception program 231.

[0084] If the waveform pattern group 160, the transmission program 131, or the reception program 231 fits within this 2 MByte size, no external flash memory is required, and therefore the size of the board mounted on the microcontroller 110 can be reduced.

[0085] In addition, in order to reduce signal processing loss due to accumulated errors caused by truncation due to fixed-point arithmetic, if it is necessary to increase the dynamic range of one sample data, the following method can be adopted. That is, a 32-bit single-precision floating-point IEEE754 format (hereinafter, floating-point format) method is adopted as the amplitude value per one sample data. In this method, the memory capacity of the waveform pattern group 160 is 2000 samples x 100 symbols x 4 multi-level modulation number x 32 bits = 3.2 MByte. Since this cannot be accommodated in the built-in flash memory 150, an external flash memory is separately prepared and installed in the microcontroller 110. Also, if the transmission program 131 or the reception program 231 does not fit in 0.8 MByte, an external flash memory is separately prepared and installed.

[0086] (Built-in SRAM170 for work) The work internal SRAM 170 stores temporary data during calculation by the CPU core 111. The work internal SRAM 170 is also a static random access memory (SRAM) that stores the transmission signal 5a that the transmission program 131 finally sends out and the reception data that the reception program finally demodulates. The work internal SRAM 170 is also used as a storage memory when the DMA controller 130 DMA transfers data from the AD converter 930 to the work internal SRAM 170. The capacity of the work internal SRAM 170 in a commercially available microcontroller 110 is assumed to be 640KByte. However, if the amount of temporary data is large and the data cannot be stored in the work internal SRAM 170, an external SRAM or DRAM (dynamic random access memory) may be used.

[0087] (DMA Controller 130) The DMA controller 130 has a continuous data transfer function and an automatic data transfer function for data in the built-in memory 120 or built-in peripherals. The DMA controller 130 performs DMA transfer when it is instructed by the CPU core 111 of a transfer source address, a transfer destination address, and the amount of data to be transferred, and is sent a start command. In other words, the DMA controller 130 enables automatic continuous data transfer from the AD converter 930 to the built-in work SRAM 170, and automatic continuous data transfer from the built-in flash memory 150 to the DA converter 610.

[0088] (DA converter 610) The DA converter 610 is a digital-to-analog converter that converts the digital signal in the DA conversion and amplification unit 6 of the first embodiment into an analog signal. The DA converter 610 of the commercially available microcontroller 110 is assumed to be as follows. The DA converter 610 continuously converts a 12-bit digital signal into an analog signal. At this time, the input load of the amplifier 620, which is an analog output load, is appropriately selected. At the same time, the DA converter 610 continuously converts the digital signal into an analog signal at a conversion rate of 200 kHz corresponding to the sampling frequency of 200 kHz described in the first embodiment, by a software trigger method or a method utilizing the DMA controller 130, which will be described later.

[0089] (A / D converter 930) The AD converter 930 is a part of the function of the amplification and AD conversion unit 9 in the first embodiment, and performs analog-to-digital conversion. The AD converter 930 is one of the built-in peripherals of the microcontroller 110, and converts an analog signal into a 12-bit digital signal.

[0090] (RTC140) The RTC 140 is one of the built-in peripherals of the microcontroller 110, and is a timer that holds values ​​of year, month, day, hour, minute, second, and subsecond. The RTC 140 counts time based on an external clock that uses, for example, a quartz crystal oscillator as a transmission source. The RTC 140 of the transmitter 1 and the RTC 140 of the receiver 2 of the opposing underwater communication system 3 that is the communication partner are set to the same time using a 1PPS (Pulse Per Second) signal that can be obtained from a GPS receiver. Then, both RTCs 140 are made to count in synchronization. The transmission program 131 and the reception program 231 generate an interrupt at 1-second intervals to the time of each RTC 140, and each obtains a time reference signal.

[0091] This concludes the description of the hardware built into the microcontroller 110. Next, we will explain the peripheral hardware of the microcontroller 110. As shown in Fig. 15, the receiver 8, the LPF / amplifier 910, the amplifier 920, the amplifier 620, and the transmitter 7 are the peripheral hardware of the microcontroller 110.

[0092] (Receiver 8) The receiver 8 is the same as the receiver in embodiment 1. Since the SMSS system uses multiple carrier waves, the frequency band tends to be wide. For this reason, it is preferable to use a hydrophone that has as uniform a sensitivity characteristic as possible across the sound wave frequency band.

[0093] (LPF / Amplifier 910) The LPF / amplifier 910 is part of the function of the amplifier / AD converter 9 in the first embodiment. The LPF / amplifier 910 is made up of a differential amplifier that amplifies the minute voltage signal from the receiver 8, and an anti-aliasing filter. The anti-aliasing filter is an LPF (Low Pass Filter) for suppressing aliasing distortion at half the sampling frequency, i.e., at or above the Nyquist frequency. In the case of a sampling frequency of 200 kHz in the communication simulation example in the first embodiment, it is configured as an analog circuit as an LPF with a cutoff frequency of 100 kHz.

[0094] (Amplifier 920) The amplifier 920 is a part of the function of the amplification and AD conversion unit 9 of the embodiment 1. The amplifier 920 adjusts the level of the input analog signal of the AD converter 930 so that the output signal of the LPF and amplifier 910 is appropriately converted into a 12-bit dynamic range which is the digital output of the AD converter 930.

[0095] (Amplifier 620) The amplifier 620 is an amplifier section with the same amplification function as the DA conversion and amplification section 6 of the first embodiment, and drives the wave transmitter 7. The amplifier 620 amplifies the analog signal output from the DA converter 610, which is a transmission signal attenuated as a sound wave due to attenuation and absorption in the underwater propagation path, to a voltage level that can be received by the opposing receiver 2. The amplifier 620 drives the wave transmitter 7 with the amplified electrical signal.

[0096] (Transmitter 7) The transmitter 7 is the same as the transmitter in the first embodiment. Since the SMSS system uses multiple carrier waves, the frequency band tends to be wide. For this reason, it is preferable to use a transducer that has as uniform an output characteristic as possible in the frequency band of the sound waves.

[0097] (Transmitter 1) The transmitter 1 includes a DA conversion and amplification unit 6, a wave transmitter 7, a CPU core 111, a DMA controller 130, a transmission program 131 in an internal flash memory 150, a waveform pattern group 160 in the internal flash memory 150, an internal working SRAM 170, and an RTC 140.

[0098] 16, the transmitter 1 includes, as program components, a data division unit 4 and a transmission waveform generation unit 5. The data division unit 4 and the transmission waveform generation unit 5 are programs that realize the same functions as the processes in the first embodiment, and are realized by a transmission program 131. The CPU core 111 executes the transmission program 131, thereby realizing the functions of the data division unit 4 and the transmission waveform generation unit 5.

[0099] (Data division part 4) In the data division unit 4, similarly to the first embodiment, the bit string of the transmission data to be transmitted is divided into divided data lengths, in this case, 2 bits which is the code length of one symbol.

[0100] (Transmission waveform generator 5) Similarly to the first embodiment, the transmission waveform generating unit 5 also uses the rank divided data [symbol number s: divided data m] as an argument to sequentially select sample data for the H number of samples from the waveform pattern group 160 of MS pieces × H sample data. Then, the transmission waveform generating unit 5 inserts I pieces of sample data with the value zero as a no-signal section, and generates a transmission signal by concatenating these waveform patterns.

[0101] (DA conversion and amplification section 6) The DA conversion and amplification unit 6 of the transmitter 1 is composed of a DA converter 610, which is one of the built-in peripherals of the microcontroller 110, and an amplifier 620 provided outside the microcontroller 110. Sample data generated with a sampling frequency resolution for one communication frame is converted from a digital signal to an analog signal by the DA converter 610, and the sample data is amplified with a period T s There are two methods for the DA converter 610 to continuously convert the sample data: a software trigger method and a method using the DMA controller 130.

[0102] (Software trigger method) In the case of the software trigger method, the transmission program 131 controls the CPU core 111. The transmission program 131 sequentially extracts one sample data piece from the waveform pattern or sample data of the non-signal section stored in the built-in memory 120 via the CPU core 111, and transfers it to the DA converter 610. Thereafter, the transmission program 131 writes a conversion trigger signal of 1 to a specified bit of a control register in the DA converter 610, thereby carrying out actual DA conversion. In the case of the communication simulation case of the first embodiment, the transmission program 131 performs the actual DA conversion by setting the sampling frequency, i.e., conversion rate, to 200 kHz, T sBy repeating this operation, the transmission program 131 converts all sample data of the transmission signal 5a into analog signals in succession and outputs them to the amplifier 620.

[0103] (Method using DMA controller 130) In the method using the DMA controller 130, the DMA controller 130 extracts a signal from the built-in memory 120 one sample data at a time, and transfers the extracted signal to the DA converter 610 continuously as the transmission signal 5a. The DMA controller 130 uses a timer (not shown), which is one of the built-in peripherals, to provide the DA converter 610 with a conversion trigger signal generated at a conversion rate of 200 kHz every time one sample data is transferred. This causes the DA converter 610 to perform continuous conversion and generate an analog signal. Furthermore, the DA converter 610 sends a request signal to the DMA controller 130 every time it converts one sample data. In response to the request signal, the DMA controller 130 transfers the sample data to be converted next to the DA converter 610. The method using the DMA controller 130 has the advantage that the CPU core 111 is free during DMA transfer, and the CPU core 111 can be used for other processing. For example, the CPU core 111 can reduce processing delays between communication frames by performing parallel operations of data division processing of the next communication frame, or the CPU core 111 can perform parallel processing of the reception program 231. In this way, the CPU core 111 can perform parallel processing of completely different application programs.

[0104] Next, a sequence when the DMA controller 130 is used will be described. The actual processing of the data division unit 4 and the transmission waveform generation unit 5 is as follows. That is, it is realized by a control program of the DMA controller 130 that controls the DMA transfer from the built-in flash memory 150 to the DA converter 610, and a control program of a timer (not shown) that generates a trigger for the conversion timing of the DA converter 610. A case will be described where the transmission signal 5a of one communication frame shown in Fig. 5 is sent to the DA converter 610 at a sampling frequency of 200 kHz, which is a communication simulation case of the first embodiment.

[0105] In the transmission program 131 of the built-in flash memory 150, a data string with the value zero is prepared in advance as I pieces of sampling data for embedding zero data in the non-signal section. The control program of the DMA controller 130 sets the number of sampling data H in the transfer data number register of the DMA controller 130, and sets the DA converter 610 in the transfer destination address register. Furthermore, the control program of the DMA controller 130 sets the start address of the waveform pattern group 160 + offset value in the transfer source address register. Here, the offset value is symbol number s × transmission code m × number of sampling data H. Initially, the symbol number s of the offset value is set to zero, the transmission code m of the first symbol is inserted, the offset value is determined, and then the start address of the waveform pattern of the first symbol is set in the transfer source address register.

[0106] Thereafter, the control program of the DMA controller 130 starts up a DMA transfer, and transfers H pieces of sampling data, which are the waveform pattern of the first symbol, to the DA converter 610. The transmission program 131 becomes aware of the transfer completion due to a transfer completion interrupt. To transfer I pieces of zero data for the no-signal interval, the control program of the DMA controller 130 then sets the start address of the I pieces of zero data for the no-signal interval in the transfer source address register, sets the number I of zero data for the no-signal interval in the transfer data count register, and sets the DA converter 610 in the transfer destination address register.

[0107] Thereafter, the control program of the DMA controller 130 starts DMA transfer and performs DMA transfer to the DA converter 610. With this operation, after the H pieces of sampling data which are the waveform pattern of the first symbol at the beginning, I pieces of zero data in a signal-free period are continuously sent to the DA converter 610. With the transfer completion interrupt, the transmission program 131 knows that the transfer of I pieces of zero data has been completed. Next, the control program of the DMA controller 130 increments the offset value s, changes m with the extracted 2-bit data, sets it in the transfer destination address register, and repeats the DMA transfer of H pieces of sampling data the number of symbols S. That is, the offset value s is changed as s=0, 1, 2...S-1, and is changed using the transmission code m of the extracted 2-bit data corresponding to the s+1th symbol. In this way, every time the DMA controller 130 transfers a waveform pattern that is H pieces of sampling data, the control program of the DMA controller 130 appropriately sets the top address of the waveform pattern that is H pieces of sampling data read from the waveform pattern group 160 in the transfer destination address register.

[0108] The control program of the DMA controller 130 is sym Each time the transfer of the waveform pattern of H sampling data corresponding to is completed, I pieces of zero data of the no-signal section are DMA-transferred to the DA converter 610 in the same procedure as the operation after the waveform pattern of the first symbol is transmitted. int Generate.

[0109] Next, a control sequence of a timer that generates a trigger for the conversion timing of the DA converter 610 will be described. This timer is not shown. The transmission program 131 sets a sampling frequency (200 kHz in the communication simulation case of the first embodiment), that is, a sampling period T sIt is set so that S×(H+I) conversion trigger signals are sent to the DA converter 610 at a period of 5 μsec (in the case of the communication simulation of the first embodiment). The sampling frequency is 200 kHz in the case of the communication simulation of the first embodiment, and T s In the communication simulation case of the first embodiment, it was 5 μsec.

[0110] Immediately after the first DMA transfer is started, the DA converter 610 sends a DMA request to the DMA controller 130, and one sample of data is transferred to the DA converter 610. Immediately after that, the timer sends the first conversion trigger, and the DA converter 610 performs digital-to-analog conversion on the transferred one sample of data. Next, the DA converter 610 sends a DMA request to the DMA controller 130 . In response to the request, the DMA controller 130 transfers the next sample data to the DA converter 610 . The timer then triggers the next conversion. s The above process repeats the transfer of sample data and digital-to-analog conversion one after another.

[0111] Here, the transmission program 131 performs the T s It is necessary to control the DMA controller 130 so that the transfer of the next sample data to the DA converter 610 is completed within this period. In particular, after the transfer completion interrupt of the DMA controller 130, it is necessary to quickly execute a series of offset value calculations, register settings, and DMA startup for the next DMA transfer. In this regard, a commercially available microcontroller 110, for example, the STM32L4R5AI, is able to fully meet the expected processing time as follows. That is, the maximum CPU clock of the STM32L4R5AI is 120MHz, and the CPU cycle time is 8.33nsec. Then, the T sFor a 1000 ms delay of 1000 ms, a series of processes must be completed within 600 CPU clock cycles. This series of processes is a combination of load / store and arithmetic instructions that require 1 to 5 clocks, but can be realized with machine language instructions consisting of at most 30 steps. Therefore, the microcontroller 110 can fully satisfy the expected specified processing time.

[0112] The amplifier 620 of the transmitter 1 is an amplifier section with the same amplification function as the DA conversion and amplification section 6 of the first embodiment, and drives the wave transmitter 7. The amplifier 620 amplifies the transmission signal, which is attenuated as a sound wave due to attenuation and absorption in the underwater propagation path, to a voltage level that can be received by the opposing receiver 2. The amplifier 620 drives the wave transmitter 7 with the amplified electrical signal. The wave transmitter 7 of the transmitter 1 is the same transducer as in the first embodiment. This SMSS method uses multiple carrier waves. For this reason, the frequency band tends to be wide, so it is preferable to use a transducer with output characteristics as uniform as possible in the frequency band of sound waves.

[0113] Next, a description will be given of the operation of the receiver 2. The receiver 2 is composed of a receiver 8, an amplifier / AD converter 9, a CPU core 111, a DMA controller 130, a reception program 231 in an internal flash memory 150, a waveform pattern group 160 in the internal flash memory 150, an internal working SRAM 170, and an RTC 140.

[0114] As shown in FIG. 16, the receiver 2 includes, as program components, a BPF 940, a signal extraction unit 13, a reference waveform generation unit 10, a plurality of correlation calculation units 11, a correlation comparison unit 12, and a data combination unit 14. The BPF 940, the signal extraction unit 13, the reference waveform generation unit 10, the correlation calculation unit 11, the correlation comparison unit 12, and the data combination unit 14 are programs that realize the same functions as the processes in the first embodiment. These functions are realized by a reception program 231. The CPU core 111 executes the reception program 231 to realize these functions.

[0115] The transmission signal 6a, which is a sound pressure signal collected by the receiver 8, is first converted to a minute voltage signal and amplified by the LPF / amplifier 910. Thereafter, in the LPF / amplifier 910, since the sampling frequency is 200 kHz, which is higher than the Nyquist frequency, in the communication simulation case of the first embodiment, the transmission signal 6a is filtered by an LPF that suppresses aliasing distortion of 100 kHz or higher. The filtered transmission signal 6a has its analog signal level adjusted by the amplifier 920, and is then input to the AD converter 930. In the second embodiment, since unnecessary waves are suppressed by a digital filter using the BPF 940 as a later-stage signal processing, the LPF does not need to be a high-cost LPF with a steep transition band.

[0116] In the AD converter 930 of the receiver 2, there are two methods for continuously converting the received signal into sample data and storing it in the built-in work SRAM 170: a software trigger method and a method using the DMA controller 130.

[0117] In the case of the software trigger method, the reception program 231 controls the CPU core 111. The reception program 231 continuously performs an operation of reading the converted data of the AD converter 930 via the CPU core 111 at a period of the sampling frequency and storing the data in the built-in work SRAM 170.

[0118] The method using the DMA controller 130 uses the DMA controller 130 and a timer (not shown) that generates a trigger for conversion timing. The DMA controller 130 continuously performs the operations of providing a conversion trigger to the AD converter 930 at a sampling period, reading the converted data of the AD converter 930 at the sampling period, and storing this converted data in the built-in work SRAM 170. The method using the DMA controller 130 has the advantage that the CPU core 111 is free during DMA transfer, making it possible for the CPU core 111 to be used for other processing. For example, by carrying out parallel operation of correlation processing of previously received signals, it is possible to reduce reception processing delays between communication frames, to process the transmission program 131 in parallel, or to process completely different application programs in parallel.

[0119] Also, in the AD converter 930 of the receiver 2, there is a method of reducing quantization errors and improving the dynamic range by using an oversampling technique. That is, in the AD conversion in the AD converter 930, the sampling frequency is set to 200 kHz, but the maximum AD conversion rate of the AD converter 930 of the STM32L4R5AI, which is the microcontroller 110, is 5 MHz. For this reason, for example, the sampling frequency is set to 3.2 MHz and sampling is performed in small increments. That is, the oversampling rate is set to 16, and moving average processing (actually, accumulation processing is performed to suppress truncation errors) of sample data sampled in small increments at 16 times is performed, and thinning processing is performed to the original sampling frequency of 200 kHz. This makes it possible to improve the S / N ratio and the dynamic range.

[0120] In general, assuming that the quantization noise is white noise and an ideal sine waveform is input at the full-scale input (peak-to-peak) of the AD converter, the SNR is expressed by Equation 6, where N is the number of digital conversion bits and k is the oversampling rate. SNR(dB)=(6.02×N)+1.76+10log(k) (Equation 6) The output bit width of the AD converter 930 in the second embodiment is assumed to be 12 bits. The ideal S / N ratio without oversampling is 74 dB. If the oversampling rate is 16, the S / N ratio is improved by 12 dB and becomes 86 dB. This is an improvement in resolution of 2 bits, which is equivalent to the resolution of a 14-bit AD converter. However, in reality, the effective resolution is smaller due to quantization noise, temperature noise, reference noise, clock jitter, etc. In this case, the 12-bit output of the AD converter 930 may grow by up to 4 bits due to cumulative processing. For this reason, the received signal is treated as sample data with a 16-bit width in the subsequent processing.

[0121] Also, instead of the AD converter 930 of the receiver 2, an external ΔΣ modulator and DFSDM, one of the built-in peripherals of the STM32L4R5AI, can be used to improve the SNR caused by quantization noise. DFSDM stands for "Digital Filter for Sigma-Delta Modulator Interface." The external ΔΣ modulator, which is the front-end part of the 1-bit ΔΣ AD converter, and the microcontroller 110 are connected through a serial interface (e.g., an SPI interface) of a 1-bit ΔΣ stream. For example, it is assumed that an external ΔΣ modulator that realizes 14 bits and 250 MSPS is selected. In this case, when decimation is performed to a sampling frequency of 200 kHz in the communication simulation case of the first embodiment, it is combined with an FIR digital filter in the DFSDM that realizes an LPF (Low Pass Filter) using an appropriate sinc function. This allows the oversampling rate to be set to 1024. Therefore, in this case, the SNR is improved by 30 dB, and the resolution is improved by 5 bits, making it equivalent to an AD converter that can obtain a resolution equivalent to 19 bits.

[0122] The BPF 940 of the receiver 2 is a finite impulse response (FIR) filter. The BPF 940 is a bandpass filter for suppressing unnecessary waves other than a necessary frequency band, as in the first embodiment. When a frequency band of 10 kHz to 30 kHz is used as in the communication simulation case of the first embodiment, the following applies. A bandpass filter that combines an HPF (High Pass Filter) with a cutoff frequency of 10 kHz and an LPF with a cutoff frequency of 30 kHz is realized by a filter having a relatively good frequency characteristic related to gain. For this purpose, for example, a digital filter having 256 taps is used, and this filter function is realized as a part of the reception program 231. The reception program 231 multiplies each of 256 sample data cut from the beginning of the "reception signal continuously sampled by the AD converter 930" stored in the work internal SRAM 170 by 256 filter coefficients in the internal flash memory 150. Then, the reception program 231 calculates the sum of the multiplication results and stores it in the work internal SRAM 170, and uses it as a new reception signal sample data string.

[0123] Next, the reception program 231 multiplies each of the 256 sample data one sample after the beginning of the received signal by the 256 filter coefficients in the built-in flash memory 150. The reception program 231 then calculates the sum of the multiplication results and stores it in the built-in work SRAM 170 to create a new received signal sample data string. From then on, the reception program 231 shifts the cut-out position from the beginning of the received signal one after another, and performs the same calculation to create a new received signal sample data string.

[0124] However, there is a shortage of sample data to be extracted at the rear of the received signal. Therefore, the reception program 231 adds zero data and performs the operation. Also, taking into account the filter delay of 128 samples of the FIR filter, the reception program 231 sets the received signal sample data string from the 128th one as a new received signal sample data string. Also, for the multiplication and summation operation, it is possible to increase the speed by using a multi-bit length fixed-point multiply-add operation instruction that the CPU core 111 has as a function. That is, the use of the UMLAL instruction. The UMLAL instruction can execute an instruction to multiply the first unsigned 32-bit operand register by the first unsigned 32-bit operand register, concatenate the 32-bit registers, and add the multiplication result to a 64-bit accumulation register in one CPU clock cycle.

[0125] 256 filter coefficients are prepared with an unsigned 32-bit width. The received signal sampled continuously by the AD converter 930 has an unsigned 12-bit width. Therefore, the multiplication result is a maximum of 44 bits, and the sum of 256 pieces is increased by 8 bits, so that the bit length of the new received signal sample data is a maximum of 52 bits. Therefore, there is no overflow beyond 64 bits. In addition, the received signal obtained by the above-mentioned oversampling method is an unsigned 16-bit. Therefore, the multiplication result is a maximum of 48 bits, and the sum of 256 pieces is increased by 8 bits, so that the bit length of the new received signal sample data is a maximum of 56 bits. Therefore, there is no overflow beyond 64 bits. The reception program 231 obtains sample data having a maximum amplitude value from the new received signal sample data string. Finally, the reception program 231 uses the maximum amplitude value to scale the new received signal sample data string as a whole, and stores it in the internal work SRAM 170 as a new received signal sample data string of, for example, 32 bits. The above is the operation of the BPF 940.

[0126] Next, the process of the signal extraction unit 13 will be described. As in the case of the signal extraction unit 13 of the first embodiment, as shown in FIG. 9, the signal extraction unit 13 extracts the maximum arrival time T c1 The signal extraction section for the second symbol and onwards is similar to the signal extraction section 13 in the first embodiment, and is set to the time position “τ1+T sym " as the standard, T sym +2T int The actual processing is realized by the correlation calculation unit 11, which will be described later, performing address calculation processing when reading out the sampling data of the received signal from the built-in SRAM 170 for work.

[0127] Next, the processing in the reference waveform generating unit 10 of the receiver 2 will be described. The reference waveform generating unit 10 executes the following processing, similar to the reference waveform generating unit 10 of the first embodiment. The reference waveform generating unit 10 extracts waveform patterns consisting of four types of H samples of sample data, [k:(00)], [k:(01)], [k:(10)], and [k:(11)], which are rank division data, corresponding to k in the kth extraction, from the waveform pattern group 160. The reference waveform generating unit 10 provides each of the extracted waveform patterns to each of the correlation calculation units 11. Note that the actual processing by the reference waveform generating unit 10 is realized by address calculation processing when reading out waveform patterns from the built-in flash memory 150 when the correlation calculation unit 11 performs correlation function calculation, which will be described later.

[0128] The waveform pattern group 160 in the built-in flash memory 150 is exactly the same as the waveform pattern group 160 held by the transmission program 131 of the transmitter 1. The waveform pattern group 160 is made up of MS × H sample waveform patterns expressed in binary format as amplitude values ​​with 12 bits per sample data.

[0129] The following is the processing content of the correlation calculation unit 11. Each correlation calculation unit 11 calculates the address of the waveform pattern to be the reference waveform by adding the top address of the waveform pattern group 160 plus offset value 1, and extracts H pieces of sample data of the required waveform pattern. Note that offset value 1 is symbol number s × transmission code m × sampling data h, where s is a variable that changes incrementally from s=1 to S, m is a variable that changes incrementally from m=1 to 4, and h is a variable that changes incrementally from h=1 to H.

[0130] In addition, each correlation calculation unit 11 performs address calculation using the start address of the received signal of one communication frame plus offset value 2 to extract H pieces of sample data within the required cut-out section of the received signal. However, when the symbol number s=1, the offset value 2 is w+sampling data h, and w ranges from 1 to T c1 From symbol number s=2 onwards, it is w + sampling data h, and w is the time position τ s +T sym ) to (T sym +2T int ) and h is a variable that changes incrementally from h = 1 to H.

[0131] In each correlation calculation unit 11, the accumulation registers A1 to A4 corresponding to the transmission code m are reset to zero as the initial value. In each correlation calculation unit 11, the offset value 1 is first calculated and set to an address indicating the symbol number s=1, the transmission code m=1, and the sample data number h=1, and the sample data is read out and used as the first multiplication operand. Next, the offset value 2 reads out sample data with symbol numbers s=1, w=1, and h=1 in the extracted section of the received signal as the second multiplication operand, multiplies the two operands, and stores the sum of these and accumulation register A1 in accumulation register A1. Next, the sample data at the address where the offset value 1 indicates the symbol number s=1, the transmission code m=2, and the sample data number h=1 is read out and set as the first multiplication operand. Then, the first multiplication operand and the second multiplication operand are multiplied together, and the result of this multiplication and the result of the accumulation register A2 are added together and stored in the accumulation register A2. Similarly, sample data corresponding to transmission code m=3 and transmission code m=4 are read out and set as the first multiplication operand, and accumulation registers A3 and A4 are calculated.

[0132] Next, the sample data number of the offset value 1 is incremented and calculated and set to an address indicating h=2, and the sample data is read out and used as the first multiplication operand. Next, the sample data number of the offset value 2 is incremented, and the sample data of h=2 is read as the second multiplication operand, the two operands are multiplied, and the result of the multiplication and the accumulation register A1 is added and stored in the accumulation register A1.

[0133] Next, the same product-sum calculation as when sample data number h=1 is performed to sequentially obtain accumulation registers A2, A3, and A4. This h variable is incremented up to H, and the accumulation registers A1, A2, A3, and A4 obtained by repeating this process are correlation function values ​​for symbol number s=1 and sample data number w=1, which correspond to transmission codes m=1 to 4.

[0134] Next, the accumulation registers A1 to A4 are reset to zero, w is incremented by 1, and the same process as above is repeated. As a result, the accumulation registers A1, A2, A3, and A4 obtain correlation function values ​​for symbol number s=1 and sample data number w=2 corresponding to transmission codes m=1 to 4. Next, increment w by one and set w = T c1 The above process is repeated until As a result, correlation function values ​​are obtained that are made up of all w pieces of sample data corresponding to symbol number s=1 and transmission codes m=4.

[0135] The process of the correlation comparator 12 will be described below. As in the first embodiment, the correlation comparator 12 calculates the PAPR from each of the correlation function value data strings corresponding to m=4 transmission codes, and identifies the received code with the highest PAPR value. As in the first embodiment, the received code is output to the data combiner 14, which performs demodulation processing to combine the multiple received codes and output the received signal 9a as received data 14a. Furthermore, as in the first embodiment, the correlation comparator 12 calculates τs, which is the peak time position of the correlation function value, and uses it to calculate an offset value 2 when extracting H sample data within the cut-out section of the received signal from symbol number s=2 onwards.

[0136] Thereafter, the calculation method for correlation function values ​​from symbol number s=2 onwards is the same as for symbol number s=1. That is, the calculation elements of offset value 1 and offset value 2 are incremented appropriately to select appropriate operands. Then, accumulation is obtained by product-sum calculation, a correlation function value data string consisting of M × w items is calculated, the received code with the highest PAPR value is identified, and the immediately preceding received code is combined to generate received data. In addition, τ s is calculated and used to calculate offset value 2 when extracting H sample data within the cut-out section of the received signal of the next symbol number.

[0137] As in the case of the BPF 940, the multiplication and summation in correlation function value calculation can be effectively performed using the UMLAL instruction, which is completed in one CPU clock cycle. The waveform data of the reference waveform pattern is a 12-bit sampling data string, and the received signal sampled continuously by the AD converter 930 is an unsigned 12-bit width, or 16 bits if oversampling is used. Therefore, the multiplication result is a maximum of 28 bits, and the summation of H=2000 samples grows by a maximum of 11 bits, so that the bit length of the correlation function value is a maximum of 39 bits, and does not exceed 64 bits and overflow.

[0138] In order to avoid the introduction of truncation errors during calculations in the BPF 940, correlation calculation unit 11, and correlation comparison unit 12, these processes may be calculated in floating-point format. Before calculations, the filter coefficient data of the BPF 940, each sample data of the received signal 9a, and each sample data of the reference waveform pattern group are converted, for example, to floating-point format to match the formats of the two. In the STM32L4R5AI microcontroller 110, the FPU (Floating Point Unit) in the CPU core 111 performs single-precision floating-point multiply-add calculations to obtain multiplication and summation, which are the main processes in the BPF 940 and correlation calculation unit 11. The FPU of this microcontroller 110 has a VFMA (Fused-Multiply-AnDAccumulate) instruction. This instruction allows floating-point multiplication and accumulation calculations to be performed in three CPU clock cycles, making it possible to perform calculations faster than microcontrollers that do not have an FPU.

[0139] The following configuration may be used as a derivative of the second embodiment. In order to reduce the processing delay between the multiplication in the BPF 940 or correlation calculation unit 11 and the product-sum calculation to obtain the sum, a dedicated product-sum calculation engine consisting of multiple multipliers and multiple adders of a fixed-point or floating-point type using an FPGA may be provided on the external bus of the microcontroller 110. By using this configuration to perform parallel calculations of the FIR digital filter of the BPF and correlation function calculations, high-speed demodulation processing is possible.

[0140] Furthermore, the microcontroller 110 may be a multi-core type microcontroller equipped with multiple CPU cores, and the correlation calculation units 11-1 to 11-4 may be configured to perform parallel arithmetic processing on each CPU core. This configuration enables high-speed demodulation processing.

[0141] As in the first embodiment, the configuration in the second embodiment also functions as a positioning system using the SSBL method by including a plurality of receivers 2 having receivers 8 arranged at different positions.

[0142] ***Effects of the second embodiment*** The effects of the second embodiment are as follows. (1) The underwater communication system 3 has the same effects as those of the first embodiment. (2) In the second embodiment, the hardware of the transmitter 1 and receiver 2 is implemented mainly around the microcontroller 110. This allows the CPU core 111, DMA controller 130, RTC 140, and waveform pattern group 160 to be shared, which has the effect of reducing the hardware scale. In addition to this effect, since the AD converter 930 and DA converter 610 are built-in, it is possible to realize a compact underwater acoustic communication modem with integrated transmission and reception, which is required for installation on underwater moving bodies such as UUVs. (3) When using the STM32L4R5AI as a microcontroller, it can be put into low-power modes such as sleep mode when it is not required to operate. This is expected to reduce problems when installed on battery-powered underwater vehicles such as UUVs.

[0143] The transmitter 1 and the receiver 2 described in the first and second embodiments can be industrially used in underwater communication devices, underwater positioning devices, underwater robots, unmanned submersibles, underwater drones, and the like. [Explanation of symbols]

[0144] 1 transmitter, 2, 2-1, 2-2, 2-3, 2-4, 2-5 receiver, 3 underwater communication system, 4 data division section, 4a transmission code, 5 transmission waveform generation section, 5a transmission signal, 6 DA conversion and amplification section, 6a transmission signal, 7 transmitter, 8 receiver, 9 amplification and AD conversion section, 9a reception signal, 10 reference waveform generation section, 10a reference signal, 11-1, 11-2, 11-3, 11-4 correlation calculation section, 12 correlation comparison section, 12a reception code, 13 signal extraction section, 14 data combination section, 16 receiver group, 31 frame, 110 microcontroller, 120 built-in memory, 111 CPU core, 130 DMA controller, 131 transmission program, 140 RTC, 150 built-in flash memory, 160 waveform pattern group, 170 Built-in SRAM for work, 231 receiving program, 610 DA converter, 620 amplifier, 910 LPF amplifier, 920 amplifier, 930 AD converter, 940 BPF.

Claims

1. An underwater communication system comprising a transmitter and a receiver, The transmitter includes: a transmission signal generating unit that generates a transmission signal, the transmission signal being a digital signal corresponding to the transmission data, the transmission signal being a digital signal in which the transmission data is divided into data pieces of the same number of bits, the waveform patterns being determined for each of the divided data pieces from their values ​​and transmission orders, and the transmission signals being arranged in the transmission order, and transmission intervals being provided between the waveform patterns; a wave transmitter that transmits the transmission signal converted into an analog signal as a wave transmission signal that is a sound wave; Equipped with The receiver includes: A receiver for receiving the transmitted signal; a signal extraction unit for extracting a waveform of each symbol from a received signal including a plurality of symbols obtained by converting the transmission signal into a digital signal, as an extracted signal, by positioning the symbol length at the center of an extraction section for each symbol; a correlation calculation unit which regards each of all bit data obtained from the same number of bits as a value of each divided data, regards the number of times the cut-out signal is cut out as the transmission order, uses a plurality of the waveform patterns determined from the value of each divided data and the transmission order as reference signals, and performs cross-correlation function calculation between the cut-out signal and each reference signal; a correlation comparator that determines, from a result of the cross-correlation function calculation, a reference signal corresponding to the cut-out signal among the reference signals, and determines a bit value indicated by the cut-out signal and a reception time of the cut-out signal based on the determined reference signal; An underwater communication system comprising:

2. The number of identical bits is 2. The underwater communication system according to claim 1, wherein the number of bits of each divided data obtained by dividing the transmission data in the transmitter into data of the same number of bits is a number of bits indicating the amount of information of a symbol to be transmitted.

3. The transmission signal generating unit of the transmitter 3. The underwater communication system according to claim 1, wherein the length of the transmission interval is adjusted in accordance with an underwater sound wave propagation environment.

4. The receiver further comprises: A reference signal adjustment unit performs an expansion / contraction conversion for expanding or contracting the waveforms of a plurality of reference signals, The correlation calculation unit is The underwater communication system according to claim 1 or 2, wherein the cross-correlation function calculation is performed using the reference signals that have been subjected to the expansion / contraction transformation.

5. The underwater communication system includes: A plurality of the receivers are provided, Each of the receivers of the multiple receivers has 3. The underwater communication system according to claim 1, wherein the wave receiver is disposed at a position different from that of the other receivers.

6. The position of each of the receivers of the plurality of receivers is 6. The underwater communication system according to claim 5, wherein the position is determined according to a positioning method for determining the position of the transmitter.

7. In a transmitter used for underwater communications, a transmission signal generating unit that generates a transmission signal, which is a digital signal corresponding to the transmission data, in which the transmission data is divided into data of the same number of bits and the waveform patterns determined for each divided data from the values ​​of the divided data and the transmission order are arranged in the transmission order, and in which a transmission interval is provided between the waveform patterns, and that adjusts the length of the transmission interval in accordance with the underwater ultrasonic propagation environment when generating the transmission signal; a wave transmitter that transmits the transmission signal converted into an analog signal as a wave transmission signal that is a sound wave; A transmitter comprising:

8. In a receiver used for underwater communications, a receiver for receiving a transmission signal obtained by converting a transmission signal, which is a digital signal corresponding to the transmission data, into an analog signal, the transmission signal being a digital signal corresponding to the transmission data, the transmission signal being converted into an analog signal, the transmission signal being arranged in a transmission order with a transmission interval between the waveform patterns being spaced apart, and the transmission data being a digital signal corresponding to the transmission data; a signal extraction unit for extracting a waveform of each symbol from a received signal including a plurality of symbols obtained by converting the transmission signal into a digital signal, as an extracted signal, by positioning the symbol length at the center of an extraction section for each symbol; a correlation calculation unit that regards each of all bit data obtained from the same number of bits as a value of each divided data, regards the number of times the cut-out signal is cut out as the transmission order, and uses a plurality of the waveform patterns determined from the value of each divided data and the transmission order as reference signals, and performs cross-correlation function calculation between the cut-out signal and each reference signal; a correlation comparator that determines, from a result of the cross-correlation function calculation, a reference signal corresponding to the cut-out signal among the reference signals, and determines a bit value indicated by the cut-out signal and a reception time of the cut-out signal based on the determined reference signal; A receiver comprising:

Citation Information

Patent Citations

  • Underwater communication system

    JP2006109279A

  • Information transmitting device using audio

    JP2011151436A

  • Underwater ultrasonic communication device using OFDM modulation having performance deterioration prevention mechanism against position fluctuation

    JP2017195566A

  • Underwater communication system and underwater communication method

    JP2020068500A

  • Acoustic communication system, acoustic transmission device, acoustic reception device, program, and acoustic signal transmission method

    JP2021196562A