Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/multidisciplinary-digital-publishing-institute/a-robust-real-time-automatic-recognition-prototype-for-maritime-400QdQwmti
References
References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.
- Publisher
- Multidisciplinary Digital Publishing Institute
- Copyright
- © 1996-2020 MDPI (Basel, Switzerland) unless otherwise stated Disclaimer The statements, opinions and data contained in the journals are solely those of the individual authors and contributors and not of the publisher and the editor(s). Terms and Conditions Privacy Policy
- ISSN
- 2076-3417
- DOI
- 10.3390/app10041227
- Publisher site
- See Article on Publisher Site
Abstract
applied sciences Article A Robust Real-Time Automatic Recognition Prototype for Maritime Optical Morse-Based Communication Employing Modified Clustering Algorithm y ,y Xiaozheng Wang , Minglun Zhang * , Hongyu Zhou, Xinglong Lin and Xiaomin Ren State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunications, Beijing 100876, China; wangduo666@bupt.edu.cn (X.W.); zhouhongyu@bupt.edu.cn (H.Z.); xllin199409@gmail.com (X.L.); xmren@bupt.edu.cn (X.R.) * Correspondence: zhangml@bupt.edu.cn y These authors contributed equally. Received: 31 December 2019; Accepted: 7 February 2020; Published: 12 February 2020 Abstract: In maritime communications, the ubiquitous Morse lamp on ships plays a significant role as one of the most common backups to radio or satellites just in case. Despite the advantages of its simplicity and eciency, the requirement of trained operators proficient in Morse code and maintaining stable sending speed pose a key challenge to this traditional manual signaling manner. To overcome these problems, an automatic system is needed to provide a partial substitute for human e ort. However, few works have focused on studying an automatic recognition scheme of maritime manually sent-like optical Morse signals. To this end, this paper makes the first attempt to design and implement a robust real-time automatic recognition prototype for onboard Morse lamps. A modified k-means clustering algorithm of machine learning is proposed to optimize the decision threshold and identify elements in Morse light signals. A systematic framework and detailed recognition algorithm procedure are presented. The feasibility of the proposed system is verified via experimental tests using a light-emitting diode (LED) array, self-designed receiver module, and microcontroller unit (MCU). Experimental results indicate that over 99% of real-time recognition accuracy is realized with a signal-to-noise ratio (SNR) greater than 5 dB, and the system can achieve good robustness under conditions with low SNR. Keywords: automatic recognition; Morse code; k-means clustering; machine learning; maritime communication; real-time; free space optical communication 1. Introduction Free space optical (FSO) communication has attracted widespread attention from both academia and industry. It has been demonstrated as a promising technology as a complementary solution to conventional radio frequency (RF), fiber optics, and microwave communications, particularly for applications with size, weight, and power restrictions. The potential advantages of FSO communication including eye safety, no spectrum licensing issues, smaller and lighter payloads, low probability of intercept, and immunity from jamming make it a natural candidate for maritime long-range communication scenarios [1–4]. Among the various modulation schemes, Morse code is still widely used as a common language between maritime vessels due to its simplicity, eciency, and small bandwidth costs. Traditionally, the ubiquitously deployed Morse lamp with LEDs, particularly on naval ships, is operated by a signalman translating the message into Morse code (dots and dashes) and manually turning the lamp “on” and “o ” in the correct sequence (on–o keying), to transmit a light signal. On the recipient ship, the blinking Appl. Sci. 2020, 10, 1227; doi:10.3390/app10041227 www.mdpi.com/journal/applsci Appl. Sci. 2020, 10, 1227 2 of 16 light signal is visually observed and directly interpreted by persons trained in the skill. However, there remains several outstanding issues to this kind of manual signaling method nowadays [5,6]. First, it would require a long training period for a skilled recipient on shipboard. In practice, Morse code relies on precise time intervals between elements while operators have diculties in maintaining an absolutely stable sending pace, the data length may vary even if it is operated by the same person, and, thus, the reliability of communication cannot be guaranteed [7]. Moreover, as it is long-term and repetitive work, the recognition accuracy is susceptible to the physical and psychological conditions of the human body, which can make them unwieldy in an emergency as well [8]. To cope with the aforementioned issues, research e orts have been made for Morse code automatic recognition in wireless communications. Previous related works are mostly based on signal processing tools, such as Kalman filters, phase-locked loops, and time–frequency analysis [8–11]. Wei et al. [9] proposed an automatic method to lock the frequency of Morse code based on the phase-locked loop circuit. Xiao et al. [10] presented an automatic reception approach for high-frequency (HF) continue-wave (CW) telegrams. A Kalman filtering algorithm combined with the support vector machine (SVM) is utilized to extract the time-domain characteristics of CW signals and deal with unstable code speed. The authors of [11] adopted the Cooley–Tukey Fast Fourier Transform (FFT) algorithm to analyze the spectrogram of a noisy audio Morse signal, and implemented a real-time decoder based on a digital signal processor. In [12], a wavelet transform-based automatic decoding method was proposed for multiplexed Morse telegraph recognition, and the feasibility of this method was analyzed via experiments. With the further advances of computers and embedded technologies, various high-performance microcontrollers have come out, facilitating the implementation of automated process control, signal recognition, data processing, and algorithm verification. Furthermore, motivated by the remarkable achievements of machine learning, some machine learning (ML)-based approaches have been proposed to model Morse signals and enhance the system robustness in a time-varying noisy environment. Wei et al. [13] provided a machine-learning method for automatic Morse signal detection. An SVM classifier, named HSVM, was proposed to be trained, based on graphical features extracted from the Morse spectrum. The authors of [14] combined signal processing and the deep learning method to construct a Morse identification system, an improved feature extraction algorithm was proposed, and the experimental results indicated a better result. Yuan et al. [15] proposed a deep learning framework on blind Morse signal detection in the wideband spectrum and achieved state-of-art performance using real-world datasets. In [16], a k-means clustering algorithm was introduced to di erentiate the elements in Morse signals after parameter extraction of the time–frequency image, and the suitability of the proposed method was investigated by simulation under di erent signal-to-noise ratio (SNR) conditions. In addition, Qu et al. [17] used this k-means clustering algorithm for dynamic threshold and Morse code recognition in a noisy environment. However, existing works have focused on Morse telegraph signals recognition in RF wireless communications, and, to the best of our knowledge, no reported work has tried to present an automatic recognition scheme and performance analysis for maritime manually sent-like Morse light signals. Besides, the ML-based k-means algorithms have not been well studied in this case as well, and there is no related real-time experimental measurement yet. To address these issues, we first propose and experimentally demonstrate an automatic recognition framework of Morse light signals in maritime optical communications by an ML-based clustering approach. Specifically, we implement a flexible FSO hardware prototype to collect real-measured Morse light signals in the physical environment for the recognition algorithm test. A modified k-means clustering is designed to compute the decision threshold for binarization and identify elements including dot–dash and space recognition in Morse light signals. To improve the decoding eciency, we make modifications to the classical k-means algorithm, and a selection sorting-assisted method has been adopted. To tackle with the emerging glitches in the binarization waveform, resulting from strong noise channels, a novel error correction scheme is presented to enhance the robustness to ambient light noises. The proposed recognition approach is implemented in a microcontroller unit (MCU), and the Appl. Sci. 2020, 10, x FOR PEER REVIEW 3 of 17 Morse code recognition in a noisy environment. However, existing works have focused on Morse telegraph signals recognition in RF wireless communications, and, to the best of our knowledge, no reported work has tried to present an automatic recognition scheme and performance analysis for maritime manually sent-like Morse light signals. Besides, the ML- based k-means algorithms have not been well studied in this case as well, and there is no related real-time experimental measurement yet. To address these issues, we first propose and experimentally demonstrate an automatic recognition framework of Morse light signals in maritime optical communications by an ML-based clustering approach. Specifically, we implement a flexible FSO hardware prototype to collect real-measured Morse light signals in the physical environment for the recognition algorithm test. A modified k-means clustering is designed to compute the decision threshold for binarization and identify elements including dot–dash and space recognition in Morse light signals. To improve the decoding efficiency, we make modifications to the classical k-means algorithm, and a selection sorting-assisted method has been adopted. To tackle with the emerging glitches in the binarization waveform, resulting from strong noise channels, a novel error correction scheme is presented to enhance the robustness to ambient light noises. The proposed recognition approach is implemented in a microcontroller unit (MCU), and the performance of the prototype is analyzed in terms of recognition accuracy and system robustness via real-time decoding results derived from the MCU. Appl. Sci. 2020, 10, 1227 3 of 16 2. Principle performance of the prototype is analyzed in terms of recognition accuracy and system robustness via 2.1. Optical Morse Signal Characterization real-time decoding results derived from the MCU. Morse code is a classical encoding scheme in telecommunications composed of a unique 2. Principle sequence of short–long dots, dashes, and spaces to represent different letters of the alphabet, 2.1. Optical Morse Signal Characterization numbers, and procedural notations. It is usually transmitted by on–off radio, light, or tones [18]. Morse code is a classical encoding scheme in telecommunications composed of a unique sequence of short–long dots, dashes, and spaces to represent di erent letters of the alphabet, numbers, and While using light as a transmitting medium, the dots and dashes are represented as procedural notations. It is usually transmitted by on–o radio, light, or tones [18]. high level (light on), and the spaces or pauses refer to low level (light off). In addition, the While using light as a transmitting medium, the dots and dashes are represented as high level Morse code depends on accurate time intervals, which correspond to the pulse width in the (light on), and the spaces or pauses refer to low level (light o ). In addition, the Morse code depends on signal waveform. The duration of a dot is regarded as a basic unit, and the dash duration is accurate time intervals, which correspond to the pulse width in the signal waveform. The duration of a th dot reeis tim regar es ded that. as a Th basic e sp unit, aces and or the timdash e intduration ervals bis etthr ween ee times dot– that. dash The es, spaces betwee or ntime letter intervals s, and between dot–dashes, between letters, and between words are 1, 3, and 5 units, respectively [19]. between words are 1, 3, and 5 units, respectively [19]. As illustrated in Figure 1, it is more As illustrated in Figure 1, it is more like the Return-to-Zero OOK (RZ-OOK) scheme in optical wireless like the Return-to-Zero OOK (RZ-OOK) scheme in optical wireless communications. Unlike communications. Unlike RF, the optical Morse-encoded signal should be real and non-negative due to RF, the optical Morse-encoded signal should be real and non-negative due to the intensity the intensity modulation/direct detection (IM/DD) mechanism [20]. modulation/direct detection (IM/DD) mechanism [20]. Figure 1. Waveform of optical Morse signal interpreting “SOS.” Figure 1. Waveform of optical Morse signal interpreting “SOS.” As mentioned above, the recognition is achieved based on the di erent time intervals and signal As mentioned above, the recognition is achieved based on the different time intervals amplitude to distinguish di erent elements. However, the automatic recognition of manually sent and signal amplitude to distinguish different elements. However, the automatic recognition Morse light signals in the real world is facing two obvious challenges. One is that the unstable typing of manually sent Morse light signals in the real world is facing two obvious challenges. One pace can lead to variations in di erent elements durations, and the other is the amplitude fluctuation resulting from ambient light noise. Assume the time intervals of dots, dashes, and spaces in unstable Morse light signals are distributed in a nearly normal fashion [21]. Let us denote D and D as the dot and dash durations samples, 1 2 respectively; then, the probability density function (PDF) is expressed as (D ) f(D) = exp( ), (1) 2 2 where D ~ N (, ), D ~ N (3, ), denotes the means, and is the scale parameter. By controlling 1 2 proper and , it is able to generate manually sent-like Morse light signals based on the Morse encoding rules as the data source to verify the designed algorithm later. A histogram of the distribution of di erent element durations is shown in Figure 2, and it can be clearly seen that the duration ratio changes. Generally, the durations of di erent elements in the two figures nearly follow a ratio of 1:3 and 1:3:5:7, respectively. In addition, additive white Gaussian noise (AWGN) is considered to be added to the raw Morse data. Figure 3 plots the signal waveform in the time domain with/without noise. It is observed that the noise will cause amplitude fluctuation, but not a ect the duration of di erent elements. In addition, these duration and amplitude variations can lead to errors for recognition. Appl. Sci. 2020, 10, x FOR PEER REVIEW 4 of 17 Appl. Sci. 2020, 10, x FOR PEER REVIEW 4 of 17 is that the unstable typing pace can lead to variations in different elements durations, and is that the unstable typing pace can lead to variations in different elements durations, and the other is the amplitude fluctuation resulting from ambient light noise. the other is the amplitude fluctuation resulting from ambient light noise. Assume the time intervals of dots, dashes, and spaces in unstable Morse light signals Assume the time intervals of dots, dashes, and spaces in unstable Morse light signals are distributed in a nearly normal fashion [21]. Let us denote D1 and D2 as the dot and dash are distributed in a nearly normal fashion [21]. Let us denote D1 and D2 as the dot and dash durations samples, respectively; then, the probability density function (PDF) is expressed as durations samples, respectively; then, the probability density function (PDF) is expressed as 1 (D) 1 (D) fD ( ) exp( ) , (1) fD ( ) exp( ) (1) 2 2 2 2 2 2 2 2 where D1 ~ N (𝜇 , 𝜎 ), D2 ~ N (3𝜇 , 𝜎 ), 𝜇 denotes the means, and 𝜎 is the scale parameter. By where D1 ~ N (𝜇 , 𝜎 ), D2 ~ N (3𝜇 , 𝜎 ), 𝜇 denotes the means, and 𝜎 is the scale parameter. By controlling proper 𝜇 and 𝜎 , it is able to generate manually sent-like Morse light signals controlling proper 𝜇 and 𝜎 , it is able to generate manually sent-like Morse light signals based on the Morse encoding rules as the data source to verify the designed algorithm later. based on the Morse encoding rules as the data source to verify the designed algorithm later. A histogram of the distribution of different element durations is shown in Figure 2, and it A histogram of the distribution of different element durations is shown in Figure 2, and it can be clearly seen that the duration ratio changes. Generally, the durations of different can be clearly seen that the duration ratio changes. Generally, the durations of different elements in the two figures nearly follow a ratio of 1:3 and 1:3:5:7, respectively. In addition, elements in the two figures nearly follow a ratio of 1:3 and 1:3:5:7, respectively. In addition, additive white Gaussian noise (AWGN) is considered to be added to the raw Morse data. additive white Gaussian noise (AWGN) is considered to be added to the raw Morse data. Figure 3 plots the signal waveform in the time domain with/without noise. It is observed Figure 3 plots the signal waveform in the time domain with/without noise. It is observed that the noise will cause amplitude fluctuation, but not affect the duration of different that the noise will cause amplitude fluctuation, but not affect the duration of different elements. In addition, these duration and amplitude variations can lead to errors for Appl. Sci. 2020, 10, 1227 4 of 16 elements. In addition, these duration and amplitude variations can lead to errors for recognition. recognition. (a) (b) (a) (b) Figure 2. Histogram of (a) space and (b) dot–dash duration distributions. Figure Figure 2. 2. H Histogram istogram o of f ( (a a) ) sp space ace a and nd ( (b b) ) dot–dash dot–dash duration duration distributions. distributions. Figure 3. Time domain of optical Morse signal with additive white Gaussian noise (AWGN). Figure 3. Time domain of optical Morse signal with additive white Gaussian noise (AWGN). Figure 3. Time domain of optical Morse signal with additive white Gaussian noise (AWGN). 2.2. Modified K-Means Clustering Algorithm The k-means algorithm is one of the most commonly used clustering methods due to its simplicity and eciency. The basic idea is to partition data samples into k clusters based on the calculated distance between each element and centroid [22–24]. As the practical FSO channels su er from many factors such as ambient light noise and multi-path dispersion [20], this method is able to optimize the decision threshold of the time-varying received signal and reduce the error rate in this noisy environment. Furthermore, as analyzed in Section 2.1, a notable clustering feature is observed in terms of the element distribution of optical Morse signals, which can be e ectively di erentiated using the k-means clustering algorithm. The detailed procedure in this case is as follows: Let X denote the sampled voltage data set, and initialize k cluster centroids C = [C , C , ::: C ,] 1 2 k randomly from X, compute the Euclidean distance between each voltage sample and centroid to find the positions of the clusters’ centroids minimizing the cost function [21], and then assign each sample to its closest cluster centroid, defined as j = arg(minkX C (m)k ), (2) 1ik Appl. Sci. 2020, 10, 1227 5 of 16 wherekk represents the Euclidean distance, C (m) is the i-th cluster centroid during the m-th iteration, X2S (m), and S (m) denotes the cluster with the centroid of C (m). Then, update the centroid of each j j j cluster by calculating the mean as follows: C (m + 1) = X, (i = 1, 2,::: , k), (3) X2S (m) where C (m + 1) is the i-th cluster centroid during the (m + 1)-th iteration, n stands for the number of i i data samples in the i-th cluster, S (m) denotes the cluster with the centroid of C (m), and X represents j i all the data samples in cluster S (m). If the updated centroids remain the same, it indicates that convergence is reached and then stops, described as abs ([C (m + 1) C (m)]) < abs([C (m) C (m 1)]) 2%, (i = 1, 2, ::: , k). (4) i i i i Otherwise, repeat the step until it reaches convergence. It is noted that there are also another enhanced clustering method and modified version of the Euclidean distance that yield good performance for high-dimensional datasets [25–27]. The data to be processed by the proposed automatic recognition algorithm include voltage value and level duration, both of which are one-dimensional datasets. Thus, for the ease of implementation in the embedded system, the simplest k-means method and Euclidean distance are chosen in this work, which can also e ectively meet the requirement. In order to adapt this algorithm to the application scenario, we propose several modifications to the clustering process to improve the recognition accuracy. The optimization of the k-means clustering algorithm is as follows: 1. Cluster Centroid Initialization: The digital filter will cause an amplitude level hopping e ect in the head and tail of the passing signal, leading to significant deviation. Thus, this part of the signal is designed to be abandoned in this case while selecting the cluster centroid during the iteration. 2. Selection Sort-Assisted (Figure 4): It can be easily known that the obtained k cluster centroids are in an unordered state. As the purpose of cluster analysis is to distinguish the signals with di erent durations in the Morse decoding, it is necessary to sort the final k centroids so that the Appl. Sci. 2020, 10, x FOR PEER REVIEW 6 of 17 device can e ectively identify di erent clusters. Considering the small quantity of centroids and easy implementation in this case, we exploit the widely used selection sort algorithm to rearrange the small quantity of centroids and easy implementation in this case, we exploit the the centroids in a certain order [28]. widely used selection sort algorithm to rearrange the centroids in a certain order [28]. 7 3 5 9 8 2 1 1 3 5 9 8 2 7 1 2 5 9 8 3 7 1 2 3 9 8 5 7 1 2 3 5 8 9 7 1 2 3 5 7 9 8 1 2 3 5 7 8 9 Figure 4. Selection sort process. Figure 4. Selection sort process. 3. System Overview A schematic diagram of the proposed system is shown in Figure 5, composed of a source computer, arbitrary waveform generator (AWG), LED driver, LED source, lens, receiver module, analog–digital converter (ADC), and MCU. First, the information bits are encoded by a PC into an unstable Morse code set and converted to an analog signal by AWG. Then, the signal is amplified and modulated to drive the LED via the driver circuit for transmitting visible light signals. At the receiver side, the incoming light signal is received by a photodetector (PD) for photoelectric conversion and amplified by a transimpedance amplifier (TIA). Then, the ADC is used to convert the electrical analog signal into a digital signal. Afterward, the digital signal is fed to the MCU for further processing. Figure 5. Block diagram of hardware setup. Appl. Sci. 2020, 10, x FOR PEER REVIEW 6 of 17 the small quantity of centroids and easy implementation in this case, we exploit the widely used selection sort algorithm to rearrange the centroids in a certain order [28]. 7 3 5 9 8 2 1 1 3 5 9 8 2 7 1 2 5 9 8 3 7 1 2 3 9 8 5 7 1 2 3 5 8 9 7 1 2 3 5 7 9 8 1 2 3 5 7 8 9 Figure 4. Selection sort process. Appl. Sci. 2020, 10, 1227 6 of 16 3. System Overview A schematic diagram of the proposed system is shown in Figure 5, composed of a source 3. System Overview computer, arbitrary waveform generator (AWG), LED driver, LED source, lens, receiver A schematic diagram of the proposed system is shown in Figure 5, composed of a source computer, module, analog–digital converter (ADC), and MCU. First, the information bits are encoded arbitrary waveform generator (AWG), LED driver, LED source, lens, receiver module, analog–digital by a PC into an unstable Morse code set and converted to an analog signal by AWG. Then, converter (ADC), and MCU. First, the information bits are encoded by a PC into an unstable Morse the signal is amplified and modulated to drive the LED via the driver circuit for transmitting code set and converted to an analog signal by AWG. Then, the signal is amplified and modulated visible light signals. At the receiver side, the incoming light signal is received by a to drive the LED via the driver circuit for transmitting visible light signals. At the receiver side, the incoming light signal is received by a photodetector (PD) for photoelectric conversion and amplified photodetector (PD) for photoelectric conversion and amplified by a transimpedance by a transimpedance amplifier (TIA). Then, the ADC is used to convert the electrical analog signal into amplifier (TIA). Then, the ADC is used to convert the electrical analog signal into a digital a digital signal. Afterward, the digital signal is fed to the MCU for further processing. signal. Afterward, the digital signal is fed to the MCU for further processing. Figure 5. Block diagram of hardware setup. Figure 5. Block diagram of hardware setup. The Cortex M4-based MCU STM32 is the core part of this whole hardware system, which has a rich peripheral interface with high integration, composed of a serial peripheral interface (SPI), finite impulse response (FIR) filter, timers (TIM), universal synchronous/asynchronous receiver transmitters (USART), and a decode module. The control signal and sampled data stream are transmitted between the MCU chip and ADC using a SPI. After FIR denoising, the sampled signal is further processed by the decode module, where the recognition algorithm is implemented. Then, the USART module is used for the interaction between the automatic recognition system and the host computer, including the transmission of control commands and return of the decoding results. In addition, the TIM module provides a working clock for the entire system to ensure that each functional component can run in an orderly manner. As indicated in Figure 6, the implementation of our proposed recognition algorithm contains three stages: Signal pretreatment, Morse decoding, and accuracy analysis. The input sampled signal is filtered first by the FIR module to reduce the e ects of noise on the subsequent process. Then, the threshold between the high level and low level is determined to classify all the voltage samples into two levels (0 and 1) in the stage of binarization. As there still exist some irregular hopping points after FIR denoising, which will a ect the automatic decoding, resulting in the decline of recognition accuracy, the error correction stage is designed as a necessary complement to address this issue by eliminating the spikes in the signal. Afterward, a relatively pure and stable Morse code signal is obtained and classified according to the characteristics of the dot, dash, and space code. After utilizing the dot–dash and space recognition algorithm, the identified Morse code set is able to be interpreted Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 17 The Cortex M4-based MCU STM32 is the core part of this whole hardware system, which has a rich peripheral interface with high integration, composed of a serial peripheral interface (SPI), finite impulse response (FIR) filter, timers (TIM), universal synchronous/asynchronous receiver transmitters (USART), and a decode module. The control signal and sampled data stream are transmitted between the MCU chip and ADC using a SPI. After FIR denoising, the sampled signal is further processed by the decode module, where the recognition algorithm is implemented. Then, the USART module is used for the interaction between the automatic recognition system and the host computer, including the transmission of control commands and return of the decoding results. In addition, the TIM module provides a working clock for the entire system to ensure that each functional component can run in an orderly manner. As indicated in Figure 6, the implementation of our proposed recognition algorithm contains three stages: Signal pretreatment, Morse decoding, and accuracy analysis. The input sampled signal is filtered first by the FIR module to reduce the effects of noise on the subsequent process. Then, the threshold between the high level and low level is determined to classify all the voltage samples into two levels (0 and 1) in the stage of binarization. As there still exist some irregular hopping points after FIR denoising, which will affect the automatic decoding, resulting in the decline of recognition accuracy, the error correction stage is designed as a necessary complement to address this issue by eliminating the spikes in the signal. Afterward, a relatively pure and stable Morse code signal is obtained and Appl. Sci. 2020, 10, 1227 7 of 16 classified according to the characteristics of the dot, dash, and space code. After utilizing the dot–dash and space recognition algorithm, the identified Morse code set is able to be into a string message sequence by table lookup decoding, and is then finally output to the computer interpreted into a string message sequence by table lookup decoding, and is then finally for accuracy computation. output to the computer for accuracy computation. Figure 6. Flowchart of optical Morse signal processing. Figure 6. Flowchart of optical Morse signal processing. 4. Signal Pretreatment 4. Signal Pretreatment The pretreatment stage includes digital filtering, binarization, and hopping points correction. The Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 17 The pretreatment stage includes digital filtering, binarization, and hopping points purpose of this stage is to provide a pure and stable Morse code signal for the subsequent decoding correction. The purpose of this stage is to provide a pure and stable Morse code signal for process. Detailed signal processing is provided in the following subsections. 4.1. Digital Filter Denoising the subsequent decoding process. Detailed signal processing is provided in the following 4.1. Digital Filter Denoising subsections. Considering that low-frequency components dominate in the frequency domain of the optical Morse signal (Figure 7), we adopt a FIR low-pass filter with a cutoff frequency of 40 Considering that low-frequency components dominate in the frequency domain of the optical Morse Hz usi signal ng th (Figur e Ham e 7), mi we ng adopt windo a FIR w to low-pass mitigat filter e th with e na oicuto se ef f fr eequency ct, and of fo40 r t Hz he using ease the of Hamming window to mitigate the noise e ect, and for the ease of implementation in the MCU as well. implementation in the MCU as well. Figure 7. Frequency domain of optical Morse signal. Figure 7. Frequency domain of optical Morse signal. 4.2. Binarization 4.2. Binarization In this stage, the proposed modified k-means clustering algorithm is utilized to compute a In this stage, the proposed modified k-means clustering algorithm is utilized to classification voltage threshold V and partition the voltage samples into two levels, representing “1” compute a classification voltage threshold VT and partition the voltage samples into two (i.e., high level) and “0” (i.e., low level), respectively. The specific steps of this process are as follows: levels, representing “1” (i.e., high level) and “0” (i.e., low level), respectively. The specific 1. Initialize a similarity value s, to quantify how similar two clusters are to one another; steps of this process are as follows: 2. Perform the mk-means algorithm to classify all the voltage data samples into two sets with the 1. Initialize a similarity value s, to quantify how similar two clusters are to one another; cluster centroids of C and C ; V1 V2 2. Perform the mk-means algorithm to classify all the voltage data samples into two sets with the cluster centroids of CV1 and CV2; 3. Calculate the distance between two centroids by d = abs (CV2 − CV1). If d < s, it means that two clusters are similar. At this time, all sampling values are divided into high or low level. Then, set VT = 0; otherwise, set VT = (CV2 + CV1)/2; 4. Iterate over all data samples; a value above the obtained threshold in step (3) indicates “1;” meanwhile, a value below indicates “0.” 4.3. Error Correction In many cases, there exist multiple unexpected glitches in signals after binarization due to the noise effects. As Figure 6 shows, the red circle marks in the block binarization indicate hopping points. Thus, we design the error correction algorithm to eliminate these abnormal hopping voltages and guarantee an accurate subsequent decoding process. The first step is to compute a threshold of durations for the purpose of localizing the emerging glitch positions. By using the algorithm in Section 2.2, duration data corresponding to each level can be partitioned into three clusters with three centroids in an ascending sort order, denoted by Cd = [Cd1, Cd2, Cd3]. Then, if Cd2/Cd1 > 3.5, it means that data samples clustered around Cd1 are from the glitches, and the basic dot duration unit is given by D0 = Cd2. Otherwise, it means the waveform is good, no jump level exists, and D0 = Cd1. Then, the duration threshold can be obtained as DT = e1*D0, where e1 denotes an adjustable parameter. In addition, we assume e1 = 0.3 at this time. The ratio 3.5 and e1 = 0.3 are established based on the experimental dataset observations for the optimal decoding result. Appl. Sci. 2020, 10, 1227 8 of 16 3. Calculate the distance between two centroids by d = abs (C C ). If d < s, it means that two V2 V1 clusters are similar. At this time, all sampling values are divided into high or low level. Then, set V = 0; otherwise, set V = (C + C )/2; T T V2 V1 4. Iterate over all data samples; a value above the obtained threshold in step (3) indicates “1;” meanwhile, a value below indicates “0.” 4.3. Error Correction In many cases, there exist multiple unexpected glitches in signals after binarization due to the noise e ects. As Figure 6 shows, the red circle marks in the block binarization indicate hopping points. Thus, we design the error correction algorithm to eliminate these abnormal hopping voltages and guarantee an accurate subsequent decoding process. The first step is to compute a threshold of durations for the purpose of localizing the emerging glitch positions. By using the algorithm in Section 2.2, duration data corresponding to each level can be partitioned into three clusters with three centroids in an ascending sort order, denoted by C = [C , d d 1 C , C ]. Then, if C /C > 3.5, it means that data samples clustered around C are from the glitches, d 2 d 3 d 2 d 1 d 1 and the basic dot duration unit is given by D = C . Otherwise, it means the waveform is good, no 0 d 2 jump level exists, and D = C . Then, the duration threshold can be obtained as D = e *D , where e 0 d 1 T 1 0 1 denotes an adjustable parameter. In addition, we assume e = 0.3 at this time. The ratio 3.5 and e = 0.3 1 1 are established based on the experimental dataset observations for the optimal decoding result. To increase the identifying accuracy, we introduce the average voltage amplitude as the second judgement criteria beyond the duration data. The average voltage amplitude refers to the statistical average of all sampled voltage values during a period of one level (“0” or “1”). Assume two new voltage thresholds, described as V = V + V e TH T pp 2 (5) V = V V e TL T pp 2 where V is the binarization voltage threshold computed in Section 4.2, V denotes the peak–peak T PP voltage value, and e represents another adjustable parameter; it is also established based on the experimental dataset observations and is set at 0.25 in this case for the optimal decoding result. In addition, using a di erent value may lead to erratic performance in our experiment. Let us denote D and V to the current duration and average voltage value during one level, c ac respectively; then, the glitches can be localized under the following constraint: D < D k V < V < V k V < V < V . (6) c ac ac T TL T T TH After marking the hopping glitches, it is time to identify their position relations to the normal voltage level. To be precise, there should be six di erent statuses of the emerging voltage spikes, which can represent all potential situations. In addition, the voltage spike can be either low or high. As long as the duration of a level is less than the computed duration threshold, it is considered as a voltage spike and merging is performed until meeting a complete voltage level (duration higher than threshold). These six cases are distinguished according to whether the transition part can form a complete level and its adjacent level condition (high or low). As the Figure 8 shows: (a) If several consecutive spikes can form a complete level after merging their duration (no longer a spike) and its adjacent level is low, then the combined one is set at high, denoted by the red line; (b) similar to (a), if the adjacent is high, then the combined one is set at low; (c) if the combined spike is still a spike after merging their duration and its adjacent level is low, then change it to low level, denoted by the red line; (d) similar to (c), if the adjacent level is high, change it to low level; (e) and (f) if the combined spike still cannot form a complete level after merging their duration, and its previous and next complete levels are of di erent types (one high and one low), then superimpose half of its duration time to the preceding and succeeding complete levels, respectively. Appl. Sci. 2020, 10, x FOR PEER REVIEW 9 of 17 To increase the identifying accuracy, we introduce the average voltage amplitude as the second judgement criteria beyond the duration data. The average voltage amplitude refers to the statistical average of all sampled voltage values during a period of one level (“0” or “1”). Assume two new voltage thresholds, described as V V V e TH T pp 2 (5) V V V e TL T pp 2 where VT is the binarization voltage threshold computed in Section 3.2, VPP denotes the peak–peak voltage value, and e2 represents another adjustable parameter; it is also established based on the experimental dataset observations and is set at 0.25 in this case for the optimal decoding result. In addition, using a different value may lead to erratic performance in our experiment. Let us denote Dc and Vac to the current duration and average voltage value during one level, respectively; then, the glitches can be localized under the following constraint: (6) Dc < DT ∥ VTL < Vac < VT ∥ VT < Vac < VTH. After marking the hopping glitches, it is time to identify their position relations to the normal voltage level. To be precise, there should be six different statuses of the emerging voltage spikes, which can represent all potential situations. In addition, the voltage spike can be either low or high. As long as the duration of a level is less than the computed duration threshold, it is considered as a voltage spike and merging is performed until meeting a complete voltage level (duration higher than threshold). These six cases are distinguished according to whether the transition part can form a complete level and its adjacent level condition (high or low). As the Figure 8 shows: (a) If several consecutive spikes can form a complete level after merging their duration (no longer a spike) and its adjacent level is low, then the combined one is set at high, denoted by the red line; (b) similar to (a), if the adjacent is high, then the combined one is set at low; (c) if the combined spike is still a spike after merging their duration and its adjacent level is low, then change it to low level, denoted by the red line; (d) similar to (c), if the adjacent level is high, change it to low level; (e) and (f) if the combined spike still cannot form a complete level after merging their duration, and its previous and Appl. next Sci. com 2020 p,le 10 t,e1227 levels are of different types (one high and one low), then superimpose hal 9 of f o 16 f its duration time to the preceding and succeeding complete levels, respectively. (a) (b) (c) (d) (e) (f) Figure 8. Error correction cases. Figure 8. Error correction cases. Take the string “SOS” as an example; Figure 9 presents its corresponding signal waveform during Take the string “SOS” as an example; Figure 9 presents its corresponding signal the whole pretreatment process. In addition, some of the error cases mentioned above are marked waveform during the whole pretreatment process. In addition, some of the error cases Appl. Sci. 2020, 10, x FOR PEER REVIEW 10 of 17 by the vertical red lines. It can be clearly seen that the emerging voltage spikes after binarization are mentioned above are marked by the vertical red lines. It can be clearly seen that the e ectively eliminated using this method, and a pure and stable Morse-encoded signal interpreting a pure and stable Morse-encoded signal interpreting “SOS” is derived after the designed emerging voltage spikes after binarization are effectively eliminated using this method, and “SOS” is derived after the designed pretreatment stage. pretreatment stage. Figure 9. Pretreatment of Morse signal interpreting “SOS”. Figure 9. Pretreatment of Morse signal interpreting “SOS.” 5. Morse Code Decoding 5. Morse Code Decoding Considering that we have obtained a relatively stable Morse-encoded signal after the pretreatment Considering that we have obtained a relatively stable Morse-encoded signal after the stage, it is easy to identify the elements using the k-means algorithm and derive the decoded data pretreatment stage, it is easy to identify the elements using the k-means algorithm and based on the Morse code table. This stage includes dot–dash recognition, space recognition, and derive the decoded data based on the Morse code table. This stage includes dot–dash accuracy computation. recognition, space recognition, and accuracy computation. 5.1. Dot–Dash Recognition 5.1. Dot–Dash Recognition Considering that both dots and dashes are represented by “1” (high level) according to Morse encoding rules, the only di erence between them is the time duration. The dot duration is the basic unit Considering that both dots and dashes are represented by “1” (high level) according to of time measurement in code transmission, while the dash duration is three units. As the time duration Morse encoding rules, the only difference between them is the time duration. The dot of the voltage level changes after pretreatment, it is necessary to recalculate the dot, as well as the duration is the basic unit of time measurement in code transmission, while the dash duration is three units. As the time duration of the voltage level changes after pretreatment, it is necessary to recalculate the dot, as well as the dash, duration thresholds. Thus, the mk- means clustering algorithm is performed to accomplish the classification of different high- level samples and dot–dash recognition. Figure 10a shows a flowchart of our dot–dash recognition algorithm: 1. Extract the high-level samples output after signal pretreatment, stored in an array; 2. Perform the mk-means method to partition these data samples into two clusters, and store the classification results in a new array HA; 3. Sort the two cluster centroids in ascending order, stored in an array CH, and ensure that CH(0) is less than CH(1). If the order of the cluster centroids changes, then update the centroid index stored in HA; 4. Assume the basic dot duration computed above is D0 and s = CH(1)/CH(0). If s < 2, it indicates that the two clusters belong to one common category. At this point, if Appl. Sci. 2020, 10, 1227 10 of 16 dash, duration thresholds. Thus, the mk-means clustering algorithm is performed to accomplish the classification of di erent high-level samples and dot–dash recognition. Figure 10a shows a flowchart of our dot–dash recognition algorithm: 1. Extract the high-level samples output after signal pretreatment, stored in an array; 2. Perform the mk-means method to partition these data samples into two clusters, and store the classification results in a new array HA; 3. Sort the two cluster centroids in ascending order, stored in an array CH, and ensure that CH(0) is less than CH(1). If the order of the cluster centroids changes, then update the centroid index stored in HA; Appl. Sci. 2020, 10, x FOR PEER REVIEW 11 of 17 4. Assume the basic dot duration computed above is D and s = CH(1)/CH(0). If s < 2, it indicates that the two clusters belong to one common category. At this point, if CH(1)/D 2, then it means CH(1)/D0 ≤ 2, then it means all the samples in HA denote “dot,” and all indexes in the all the samples in HA denote “dot,” and all indexes in the array need to be updated to 0; else, if array need to be updated to 0; else, if CH(1)/D0 > 2, then it means HA represents “dash,” CH(1)/D > 2, then it means HA represents “dash,” and the indexes should be updated to 1. and the indexes should be updated to 1. START START High-Level Low-Level Samples Samples Extraction Extraction Perform K-Means Algorithm value=dataLow(i)/D Sort the two centroids value<2? N LA(i)=1 Centroids order change ? Update HA value<4? LA(i)=3 CH(1)/CH(0) <2? value<6? N LA(i)=5 CH(1)/D LA(i)=7 >2? Iterarion Over? Update HA to 1 Update HA to 0 END END (a) (b) Figure 10. Flowchart of (a) dot–dash, and (b) space recognition algorithms. Figure 10. Flowchart of (a) dot–dash, and (b) space recognition algorithms. 5.2. Space Recognition 5.2. Space Recognition In Morse coding rules, space refers to a period of signal absence following each dot or dash, which In Morse coding rules, space refers to a period of signal absence following each dot or represents 0 (low level) in binary optical signals. Spaces between adjacent dots and dashes, two letters, dash, which represents 0 (low level) in binary optical signals. Spaces between adjacent dots words, and sentences are 1, 3, 5, and 7 units long, respectively. Therefore, a similar concept is used to and dashes, two letters, words, and sentences are 1, 3, 5, and 7 units long, respectively. classify all the low-level samples into these four statuses by calculating the time duration threshold. Therefore, a similar concept is used to classify all the low-level samples into these four Assume D = CH(0) is the new computed basic duration, used to initialize the threshold, and it is statuses by calculating the time duration threshold. Assume Dn = CH(0) is the new computed noted that while two clusters both indicate “dash” at the same time, D is supposed to be 1/3 of the smaller basic du cluster ration centr , use oid d to value. initiAs aliz illustrated e the threin shFigur old, e an 10 d b, it the is n space oted rt ecognition hat whiler tw esults o clar ust e stor ers ed bot in h indicate “dash” at the same time, Dn is supposed to be 1/3 of the smaller cluster centroid value. As illustrated in Figure 10b, the space recognition results are stored in an array LA, where the element value indicates the exact time duration units. For example, 3 denotes 3 units, representing spaces between adjacent letters. 5.3. Accuracy Computation To quantify the accuracy of our recognition algorithm, the Levenshtein distance (edit distance) is used by counting the minimum number of edit operations required to transform one string into the other. The modification operations include: (a) Insert a character; (b) delete one character; and (c) replace one character with another. Let us denote R = [r1 r2…ri] and D = [d1 d2…dj] as the raw data string and the decoded data string, respectively. m[i,j] Appl. Sci. 2020, 10, 1227 11 of 16 an array LA, where the element value indicates the exact time duration units. For example, 3 denotes 3 units, representing spaces between adjacent letters. 5.3. Accuracy Computation To quantify the accuracy of our recognition algorithm, the Levenshtein distance (edit distance) is used by counting the minimum number of edit operations required to transform one string into the other. The modification operations include: (a) Insert a character; (b) delete one character; and (c) replace one character with another. Let us denote R = [r r ::: r ] and D = [d d ::: d ] as the raw data 1 2 i 1 2 j string and the decoded data string, respectively. m[i,j] represents the edit distance between the first i characters of R and the first j characters of D, and is expressed as [29] 0 i = 0, j = 0 j i = 0, j > 0 m(i, j) = , (7) i i > 0, j = 0 min(m(i 1, j) + 1, m(i, j 1) + 1, m(i 1, j 1) + f lag) i > 0, j > 0 where the flag variable is the indicator function equal to 0 when R[i] = D[j], and equal to 1 otherwise. Then, we have the recognition accuracy, m(i, j) A = 1 , (8) max(jRj,jDj) where |R| and |D| represent the length of the raw and decoded data string, respectively. In addition, the procedure of this dynamic programming algorithm to derive the edit distance is shown in Algorithm 1: Algorithm 1 String similarity algorithm 1: procedure Levenshtein Distance (R, D) 2: int m[i,j] = 0 3: for each i 1 to |R| do 4: m[i,0] = i 5: end for 6: for each j 1 to |D| do 7: m[0, j] = j 8: end for 9: for each i 1 to |R| do 10: for each j 1 to |D| do 11: t if (R[i] = D[j]) then 0 else if 1 12: m[i,j] = min{ m[i 1,j 1] + t, m[i 1,j] + 1, m[i,j 1] + 1} 13: end for 14: end for 15: return m[|R|, |D|] 16: end procedure 6. Experimental Validation 6.1. Experimental Setup The experimental testbed for our proposed system prototype is shown in Figure 11; we use this prototype to generate a real optical Morse encoded signal and verify the feasibility of our proposed automatic recognition approach. Appl. Sci. 2020, 10, x FOR PEER REVIEW 12 of 17 represents the edit distance between the first i characters of R and the first j characters of D, and is expressed as [29] 00ij ==,0 j ij => 0, 0 , mi (, j) = (7) ii>= 0,j 0 min(mi(1−+ , j) 1,m(i, j−1)+1,m(1 i− , j−1)+ flag) i>0, j>0 where the flag variable is the indicator function equal to 0 when R[i] = D[j], and equal to 1 otherwise. Then, we have the recognition accuracy, mi (, j) A=− 1 , (8) max(RD , ) where |R| and |D| represent the length of the raw and decoded data string, respectively. In addition, the procedure of this dynamic programming algorithm to derive the edit distance is shown in Algorithm 1: Algorithm 1 String similarity algorithm 1: procedure Levenshtein Distance (R, D) 2: int m[i,j] = 0 3: for each i ←1 to |R| do 4: m[i,0] = i 5: end for 6: for each j ←1 to |D| do 7: m[0 , j] = j 8: end for 9: for each i ←1 to |R| do 10: for each j ←1 to |D| do 11: t ← if (R[i] = D[j]) then 0 else if 1 12: m[i,j] = min{ m[i-1,j-1] + t, m[i-1,j] + 1, m[i,j-1] + 1} 13: end for 14: end for 15: return m[|R|, |D|] 16: end procedure 6. Experimental Validation 6.1. Experimental Setup The experimental testbed for our proposed system prototype is shown in Figure 11; we Appl. use this prototype to generate a real optical Sci. 2020, 10, 1227 Morse encoded signal and verify the feasibility 12 of 16 of our proposed automatic recognition approach. (a) (b) Figure 11. Demonstration of proposed free space optical (FSO) prototype: (a) Transmitter side, (b) receiver side. At Tx, an unstable Morse encoded signal is first produced by an arbitrary waveform generator (AWG 7051). Then, it is modulated to the LED (XLamp XP-L2) driven by the self-designed driving circuit packed inside the lamp. Note that a signal lamp consisting of LED arrays is employed in our system for the purpose of simulating the maritime optical communication scenarios. At Rx, we use an optical lens (ZLKC-KM5012MP8) with a focal length of 50 mm to narrow the field-of view (FOV) of the receiver, thereby mitigating the incoming ambient light noise. The emitted light signal is captured by a self-designed optical receiving module including an avalanche photo-diode (S8664-50K) and a transimpedance amplifier (LTC6268-10) with a high gain factor of 2 million times. Then, after analog–digital conversion by ADS8866 (16-bit resolution), the signal is applied to the STM32F446 embedded system [30] and processed based on our proposed approach. The sampling period is T = 2.7 ms, and the sampling rate of the ADC is set at F = 375 Hz to guarantee that there are at least 20 C30 samples during the period of a basic dot unit. Finally, the output recognition results are analyzed on the computer in real time using the algorithm mentioned in Section 4.3, and the accuracy performance is determined. The key parameters of the devices used in this system are listed in Table 1. Table 1. System configuration parameters. Device Manufacturer Model AWG Tektronix AWG 7051 LED Cree Inc XLamp XP-L2 Lens ZhiSai Technology ZLKC-KM5012MP8 APD Hamamatsu Photonics S8664-50K TIA Linear Technology LTC6268-10 ADC Texas Instruments ADS 8866 MCU STMicroelectronics STM32F446 6.2. Recognition Accuracy Evaluation In the experiments, a Morse-encoded signal interpreting the string “hello world” is generated on the computer with MATLAB using the method in Section 2.1, and the durations and amplitude fluctuation are considered by controlling the and and adding AWGN, respectively. We first investigate the oine performance. Figure 12 shows the o -line pretreatment results based on the mk-means clustering algorithm. One can observe that the obtained waveform is just the same as the original signal, which indicates that our method works well. Appl. Sci. 2020, 10, 1227 13 of 16 Appl. Sci. 2020, 10, x FOR PEER REVIEW 14 of 17 Appl. Sci. 2020, 10, x FOR PEER REVIEW 14 of 17 M orse signal w ith noises M orse signal w ith noises -2060 200 123456789 Time (s) -20 A fter FIR D enoising 0 123456789 Time (s) A fter FIR D enoising 0 123456789 Time (s) A fter B in arization 0 123456789 Time (s) 0.5 A fter B in arization 0 123456789 0.5 Time (s) 0 A fter E rro r C o rrectio n 10 123456789 0.5 Time (s) A fter E rro r C o rrectio n 0 1 23 45 67 89 Time (s) 0.5 Raw M orse Data 0 1 23 45 67 89 0.5 Time (s) Raw M orse Data 0 1 23 45 67 89 Time (s) 0.5 0 1 23 45 67 89 Figure 12. Simulation results of Morse signal interpreting “hello world.” Time (s) Figure 12. Simulation results of Morse signal interpreting “hello world.” Figure 12. Simulation results of Morse signal interpreting “hello world.” Then, the prototype is used to test the real-time decoding accuracy. The optical signal emitted from LED arrays is received by the photo detector and then processed in the Then, the prototype is used to test the real-time decoding accuracy. The optical signal emitted Then, the prototype is used to test the real-time decoding accuracy. The optical signal embedded system; the translation results are obtained by the serial port software on the from LED arrays is received by the photo detector and then processed in the embedded system; the computer. emitted from LED arrays is received by th Figure 13 displays the real-time ree p cognition results in the hoto detector and then processed in the Serial Assistant translation results are obtained by the serial port software on the computer. Figure 13 displays the software derived from STM32. It is also observed that the decoding results are exactly the embedded system; the translation results are obtained by the serial port software on the real-time recognition results in the Serial Assistant software derived from STM32. It is also observed same as the raw string “hello world.” It is worth noticing that the transceiver was originally computer. Figure 13 displays the real-time recognition results in the Serial Assistant that the decoding results are exactly the same as the raw string “hello world.” It is worth noticing designed for the long-range outdoor environment and can support 4.8 km FSO software derived from STM32. It is also observed that the decoding results are exactly the that the transceiver was originally designed for the long-range outdoor environment and can support communication [31]. The transmitter consists of more than 30 LEDs, each at a DC forward 4.8 km FSO communication [31]. The transmitter consists of more than 30 LEDs, each at a DC forward same as the raw string “hello world.” It is worth noticing that the transceiver was originally current of 1500 mA and a corresponding electrical power of 4.35 W, which is enough for current of 1500 mA and a corresponding electrical power of 4.35 W, which is enough for indoor optical designed for the long-range outdoor environment and can support 4.8 km FSO indoor optical wireless communication. Thus, the received optical power maintains a high wireless communication. Thus, the received optical power maintains a high level in the laboratory communication [31]. The transmitter consists of more than 30 LEDs, each at a DC forward level in the laboratory environment, and the recognition accuracy also remains the same environment, and the recognition accuracy also remains the same after increasing the distance from 1 current of 1500 mA and a corresponding electrical power of 4.35 W, which is enough for after increasing the distance from 1 to 6 m. to 6 m. indoor optical wireless communication. Thus, the received optical power maintains a high level in the laboratory environment, and the recognition accuracy also remains the same after increasing the distance from 1 to 6 m. Figure 13. Figure 13. Real Real-time -time opti optical cal MoMorse rse code code recor gnition resu ecognition lts results. . 6.3. System Robustness Evaluation 6.3. System Robustness Evaluation To evaluate the performance and robustness of our system, we investigate the recognition accuracy To evaluate the performance and Figure 13. Real-time opti robustness cal Morse co of our system, we i de recognition results. nvestigate the under di erent signal-to-noise ratio (SNR) conditions. Considering that it is dicult to accurately recognition accuracy under different signal-to-noise ratio (SNR) conditions. Considering control the SNR of the input signal over the optical path, the experiment is conducted between the 6.3. System Robustness Evaluation that it is difficult to accurately control the SNR of the input signal over the optical path, the PC and STM32 through the serial port. During the test, two English texts containing 2196 characters experiment is conducted between the PC and STM32 through the serial port. During the were To evaluate the performance and selected as the raw data, and each character robustness appeared of our system, we i randomly. Note that the transmitting nvestigate t data he test, two English texts containing 2196 characters were selected as the raw data, and each volume is far more than that of usual use in our considered real maritime scenario. recognition accuracy under different signal-to-noise ratio (SNR) conditions. Considering The specific test procedure is as follows: First, the raw data of length 2196 characters are that it is difficult to accurately control the SNR of the input signal over the optical path, the Morse-encoded and superimposed with noise in MATLAB. By controlling the proportion of noise experiment is conducted between the PC and STM32 through the serial port. During the samples, the Morse-encoded signal data with di erent SNRs ( 3–6 dB) is generated on the PC and test, two English texts containing 2196 characters were selected as the raw data, and each Amplitude Amplitude Amplitude Amplitude Amplitude Amplitude Amplitude (mV) (mV) Amplitude Amplitude Amplitude (mV) (mV) Appl. Sci. 2020, 10, x FOR PEER REVIEW 15 of 17 character appeared randomly. Note that the transmitting data volume is far more than that of usual use in our considered real maritime scenario. The specific test procedure is as follows: First, the raw data of length 2196 characters are Morse-encoded and superimposed with noise in MATLAB. By controlling the proportion of noise samples, the Morse-encoded signal data with different SNRs (–3–6 dB) Appl. Sci. 2020, 10, 1227 14 of 16 is generated on the PC and then transmitted to the STM32-embedded system through the serial port; secondly, STM32 stores and decodes the received data and returns the recognition results back to the host computer. Finally, in the host computer, the original and then transmitted to the STM32-embedded system through the serial port; secondly, STM32 stores and decoded data file are examined, and the decoding accuracy is statistically analyzed. The decodes the received data and returns the recognition results back to the host computer. Finally, in statistical decoding results during the experiment are listed in Table 2. the host computer, the original and decoded data file are examined, and the decoding accuracy is statistically analyzed. The statistical decoding results during the experiment are listed in Table 2. Table 2. Recognition results under different signal-to-noise ratios (SNRs). Table 2. Recognition results under di erent signal-to-noise ratios (SNRs). SNR (dB) Total Characters Correct Characters Accuracy SNR (dB) Total Characters Correct Characters Accuracy −3 2196 1979 90.1% −1 2196 2003 91.2% 3 2196 1979 90.1% 11 2 2196196 20032038 91.2%92.8% 1 2196 2038 92.8% 3 2196 2102 95.7% 3 2196 2102 95.7% 5 2196 2128 96.9% 5 2196 2128 96.9% 6 2196 2176 99.1% 6 2196 2176 99.1% The curve of recognition accuracy versus SNR derived from Table 2 is plotted in Figure 14. It is noticed that the decoding accuracy increases with the growth of the SNR. Our The curve of recognition accuracy versus SNR derived from Table 2 is plotted in Figure 14. It is proposed approach can achieve an average automatic recognition accuracy of more than noticed that the decoding accuracy increases with the growth of the SNR. Our proposed approach can achieve 90% wh an en average the SNR of t automatic he input sig recognitionnal is great accuracy ofe mor r than e than−390% dB. Furthermore, an when the SNR of theaccuracy of input signal is over 95% is observed with an SNR g greater than 3 dB. Furthermore, an accuracy reater than of over 3 dB. In the 95% is observed case that the with an noise is relatively SNR greater than 3 dB. In the case that the noise is relatively strong with an SNR of 3 dB, the system can still achieve a strong with an SNR of −3 dB, the system can still achieve a good accuracy of 90.1%. Thus, good accuracy of 90.1%. Thus, we conclude that, in the presence of ambient light noise, the proposed we conclude that, in the presence of ambient light noise, the proposed mk-means based mk-means based recognition system still performs well in terms of the decoding accuracy performance. recognition system still performs well in terms of the decoding accuracy performance. Recognition accuracy under different SNR 99.10% 100.00% 96.90% 95.70% 92.80% 95.00% 91.20% 90.10% 90.00% 85.00% -3 -1 1356 SNR (dB) Figure 14. Recognition accuracy under different SNRs. Figure 14. Recognition accuracy under di erent SNRs. 7. Conclusions 7. Conclusions In conclusion, we proposed an automatic recognition scheme for maritime optical Morse lamps In conclusion, we proposed an automatic recognition scheme for maritime optical based on a modified k-means clustering algorithm. A flexible FSO communication prototype consisted Morse lamps based on a modified k-means clustering algorithm. A flexible FSO of LED arrays, and an MCU was constructed for real data collection and real-time decoding. We also communication prototype consisted of LED arrays, and an MCU was constructed for real investigated the performance of the proposed approach including recognition accuracy, as well as data collection and real-time decoding. We also investigated the performance of the system robustness. Real-time recognition results derived from the MCU indicate that the accuracy increases with the growth of the SNR, and can reach more than 99%. It is worth mentioning that our proposed approach including recognition accuracy, as well as system robustness. Real-time proposed modified k-means clustering algorithm can also be applied in other digital communication recognition results derived from the MCU indicate that the accuracy increases with the systems for decision threshold optimization and is adaptive to di erent channel characteristics. growth of the SNR, and can reach more than 99%. It is worth mentioning that our proposed In the future, we will further investigate the recognition performance based on other ML-based clustering algorithms. Author Contributions: Conceptualization, X.W. and M.Z.; methodology, X.W. and M.Z.; software, X.W., M.Z., H.Z. and X.L.; validation, X.L.; formal analysis, X.W. and M.Z.; investigation, X.W. and M.Z.; resources, H.Z. and X.L.; data curation, X.W. and M.Z.; writing—original draft preparation, X.W.; writing—review and editing, M.Z.; visualization, X.W., H.Z. and X.L.; supervision, M.Z. and X.R.; project administration, M.Z. All authors have read and agreed to the published version of the manuscript. Accuracy Appl. Sci. 2020, 10, 1227 15 of 16 Funding: This research received no external funding. Conflicts of Interest: The authors declare no conflict of interest. References 1. Khalighi, M.A.; Uysal, M. Survey on free space optical communication: A communication theory perspective. IEEE Commun. Surv. Tutor. 2014, 16, 2231–2258. [CrossRef] 2. Grant, K.J.; Mudge, K.A.; Clare, B.A.; Perejma, A.S.; Martinsen, W.M. Maritime Laser Communications Trial 98152-19703, DSTO; Endinburgh: South Australia, Australia, 2012. 3. Sluz, J.E.; Riggins, J., II; Juarez, J.C.; Sova, R.M.; Young, D.W.; Nelson, C. Characterization of data transmission through a maritime free-space optical channel with a custom bit error rate tester. Proc. Spie 2010, 7700, 77000D. 4. Yu, J.; Li, Q.; Li, H.; Wang, Q.; Zhou, G.; He, D.; Xu, S.; Xia, Y.; Huang, Y. High-Precision Light Spot Position Detection in Low SNR Condition Based on Quadrant Detector. Appl. Sci. 2019, 9, 1299. [CrossRef] 5. New Atlas. Available online: https://newatlas.com/us-navy-signal-lamps-fltc-texting/50523/ (accessed on 10 October 2019). 6. Sapaico, L.R.; Sato, M. Analysis of vision-based text entry using Morse code generated by tongue gestures. In Proceedings of the 2011 4th International Conference on Human System Interactions (HSI), Yokohama, Japan, 19–21 May 2011; pp. 158–164. 7. Hsieh, M.C.; Luo, C.H.; Mao, C.W. Unstable Morse code recognition with adaptive variable-ratio threshold prediction for physically disabled persons. IEEE Trans. Rehabil. Eng. 2000, 8, 405–413. [CrossRef] [PubMed] 8. Popular Mechanics. Available online: https://www.popularmechanics.com/military/navy-ships/a27391/us- navy-Morse-code-software/ (accessed on 10 October 2019). 9. Ma, W.; Zhang, J.; Wang, H. Automatic Decoding System of Morse Code. O.I. Autom. 2007, 26, 51–52. 10. Xiao-Na, Z.; Guo-Jun, L.; Yong, J.; Li, Z. Automatic reception of High-frequency CW telegraph with Support Vector Machine. In Proceedings of the 2nd IEEE International Conference on Future Computer and Communication, Wuhan, China, 21–24 May 2010; pp. 335–338. 11. Zahradnik, P.; Šimák, B. Implementation of Morse decoder on the TMS320C6748 DSP development kit. In Proceedings of the 2014 6th IEEE European Embedded Design in Education and Research Conference, Milan, Italy, 11–12 September 2014; pp. 128–131. 12. Xiao, S.; Gao, Y. Multiplexed Morse Telegraph Automatic Decoding Based on Wavelet Transform. Comput. Digit. Eng. 2017, 45, 632–636. 13. Wei, Z.; Jia, K.; Sun, Z. An Automatic Detection Method for Morse Signal Based on Machine Learning. In Proceedings of the 12th International Conference on Intelligent Information Hiding and Multimedia Signal, Matsue, Japan, 12–15 August 2017; pp. 185–191. 14. Wang, X.; Zhao, Q.; Ma, C.; Xiong, J. Automatic Morse Code Recognition Under Low SNR. In Proceedings of the 2018 International Conference on Mechanical, Electronic, Control and Automation Engineering, Manila, Philippines, 5–7 January 2018. 15. Yuan, Y.; Sun, Z.; Wei, Z.; Jia, K. DeepMorse: A Deep Convolutional Learning Method for Blind Morse Signal Detection in Wideband Wireless Spectrum. IEEE Access 2019, 7, 80577–80587. [CrossRef] 16. Wang, Y.; Sun, Z.; Jia, K. An Automatic Decoding Method for Morse Signal based on Clustering Algorithm. In Proceedings of the 12th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, Kaohsiung, China, 21–23 November 2016; pp. 235–242. 17. Qu, S.; Liu, H.; Zhang, X. Morse Recognition Algorithm Based on K-means. In Proceedings of the 2019 IEEE Cross Strait Quad-Regional Radio Science and Wireless Technology Conference, Taiyuan, China, 19–22 July 2019; pp. 1–2. 18. Lee, C.T.; Shen, T.C.; Lee, W.D. A Novel Optical Morse Code-Based Electronic Lock Using the Ambient Light Sensor and Fuzzy Controller. Appl. Sci. 2017, 7, 140. [CrossRef] 19. Bilal, M.; Liu, S.; Qiao, G.; Wan, L.; Tao, Y. Bionic Morse Coding Mimicking Humpback Whale Song for Covert Underwater Communication. Appl. Sci. 2020, 10, 186. [CrossRef] 20. Ma, S.; Dai, J.; Lu, S.; Li, H.; Zhang, H.; Du, C.; Li, S. Signal Demodulation with Machine Learning Methods for Physical Layer Visible Light Communications: Prototype Platform, Open Dataset, and Algorithms. IEEE Access 2019, 7, 30588–30598. [CrossRef] Appl. Sci. 2020, 10, 1227 16 of 16 21. Zhang, X. The Distinction of Ship’s Communication Signals. J. Jimei. Navig. Inst. 1996, 14, 19–25. 22. Zhang, J.; Chen, W.; Gao, M.; Shen, G. K-means-clustering-based fiber nonlinearity equalization techniques for 64-QAM coherent optical communication system. Opt. Express 2017, 25, 27570–27580. [CrossRef] [PubMed] 23. Peng, Y.; Liu, X.; Shen, C.; Huang, H.; Zhao, D.; Cao, H.; Guo, X. An Improved Optical Flow Algorithm Based on Mask-R-CNN and K-Means for Velocity Calculation. Appl. Sci. 2019, 9, 2808. [CrossRef] 24. Hou, X. An Improved K-means Clustering Algorithm Based on Hadoop Platform. In Proceedings of the International Conference on Cyber Security Intelligence and Analytics, Shenyang, China, 21–22 February 2019; pp. 1101–1109. 25. Cardarilli, G.C.; Di Nunzio, L.; Fazzolari, R.; Re, M.; Spanó, S. AW-SOM, an algorithm for high-speed learning in hardware self-organizing maps. IEEE Trans. Circuits Syst. IiExp. Briefs 2019, 67, 380–384. [CrossRef] 26. Cardarilli, G.C.; Di Nunzio, L.; Fazzolari, R.; Nannarelli, A.; Re, M.; Spanò, S. N-Dimensional Approximation of Euclidean Distance. IEEE Trans. Circuits Syst. IiExp. Briefs 2019, 55, 161. [CrossRef] 27. Tolentino, J.A.; Gerardo, B.D.; Medina, R.P. Enhanced Manhattan-Based Clustering Using Fuzzy C-Means Algorithm. In Proceedings of the International Conference on Computing and Information Technology, Chiangmai, Thailand, 5–6 July 2018; pp. 126–134. 28. Heineman, G.T.; Pollice, G.; Selkow, S. Algorithms in a Nutshell: A Practical Guide, 2rd ed.; O’Reilly Media, Inc.: Sebastopol, CA, USA, 2016; pp. 53–89. 29. Zhao, Z.; Yin, Z.; Wang, Q.; Xu, X.; Jiang, H. An improved algorithm of Levenshtein Distance and its application in data processing. J. Comput. Applicat. 2009, 29, 424–426. [CrossRef] 30. STMicroelectronics 32-bit Arm Cortex MCUs. Available online: https://www.st.com/en/microcontrollers- microprocessors/stm32f446.html (accessed on 26 January 2020). 31. Zhou, H.; Zhang, M.; Wang, X.; Ren, X. Implementation of High Gain Optical Receiver with the Large Photosensitive Area in Visible Light Communication. In Proceedings of the Asia Communications and Photonics Conference, Chengdu, China, 2–5 November 2019; pp. 99–101. © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Journal
Applied Sciences – Multidisciplinary Digital Publishing Institute
Published: Feb 12, 2020