Beza Negash Getu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Beza Negash Getu has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Hussein Attia
King Fahd University of Petroleum & Minerals (KFUPM)
6 shared publications J.B. Andersen
Dept. of Commun. Technol., Aalborg Univ., Denmark
4 shared publicationsAbdulhadi Brazi
American University of Ras Al Khaimah
1 shared publicationAbdulla Alzaabi
Department of Electrical, Electronics and Communications Engineering American University of Ras Al Khaimah, Ras Al Khaimah, UAE
1 shared publicationAbdullah Asaad
American University of Ras Al Khaimah
1 shared publicationAhmed Abbas
American University of Ras Al Khaimah
1 shared publication
Latest Publications
- Open Access
Time–Frequency-Based CNN Waveform Classification: Comparison of Time-Domain, FFT, STFT, Wavelet and Hybrid Deep Learning
Reliable classification of noisy waveforms is an important problem in modern signal processing, with applications in communications, sensing, monitoring and intelligent signal analysis. Although deep learning, particularly convolutional neural networks (CNNs), has demonstrated strong capability for learning discriminative representations directly from raw sampled time domain signals, classification performance depends not only on network architecture but also on how the signal is represented before deep learning. Time-domain samples preserve complete temporal information but may provide limited explicit information about spectral structure, whereas Fourier- and time–frequency-based representations expose complementary characteristics that may be advantageous under noisy waveform conditions. This study extends our previous investigation of neural network-based waveform classification in additive white Gaussian noise (AWGN) by systematically evaluating the effect of signal representation using a controlled convolutional-learning framework. This study considers five classification configurations: time-domain CNN, FFT-based CNN, STFT-based CNN, wavelet-based CNN and a hybrid decision-level fusion model that combines the probabilistic outputs of the individual classifiers. To ensure a controlled comparison, the principal CNN architecture and training strategy are kept consistent across comparable representations, while only the representation and corresponding input dimensionality are changed. This study evaluates classification accuracy and confusion matrices across multiple SNR conditions. It is found that, at low SNR, and in many of the cases, the hybrid, STFT and wavelet are outperforming the time-based and FFT-based classifier, while all of them perform well at high SNRs. At an SNR of −5 dB, the accuracy performances are 78%, 62.1%, 84.35%, 82% and 90.1% for the time-domain, FFT, STFT, wavelet and the hybrid CNN classifier, respectively, for the test samples considered in this study. The hybrid deep learning that is based on the fusion of the four (time domain, FFT, STFT and wavelet) is always the best in terms of performance. When there is random amplitude, frequency and phase variations in the waveforms, the corresponding accuracy performances at SNR = −5 dB become 68.15%, 49.05%, 78.35%, 80.85% and 89.75% respectively, showing a decrease inaccuracy, but with hybrid still performing well exploiting the advantages from all representations. The results of this experiment are intended to explore the time–frequency representation-level contributions in deep-learning-based waveform classification rather than attributing performance improvements solely to increasingly complex network architectures.
Open AccessTime–Frequency-Based CNN Waveform Classification: Comparison of Time-Domain, FFT, STFT, Wavelet and Hybrid Deep Learning
Reliable classification of noisy waveforms is an important problem in modern signal processing, with applications in communications, sensing, monitoring and intelligent signal analysis. Although deep learning, particularly convolutional neural networks (CNNs), has demonstrated strong capability for learning discriminative representations directly from raw sampled time domain signals, classification performance depends not only on network architecture but also on how the signal is represented before deep learning. Time-domain samples preserve complete temporal information but may provide limited explicit information about spectral structure, whereas Fourier- and time–frequency-based representations expose complementary characteristics that may be advantageous under noisy waveform conditions. This study extends our previous investigation of neural network-based waveform classification in additive white Gaussian noise (AWGN) by systematically evaluating the effect of signal representation using a controlled convolutional-learning framework. This study considers five classification configurations: time-domain CNN, FFT-based CNN, STFT-based CNN, wavelet-based CNN and a hybrid decision-level fusion model that combines the probabilistic outputs of the individual classifiers. To ensure a controlled comparison, the principal CNN architecture and training strategy are kept consistent across comparable representations, while only the representation and corresponding input dimensionality are changed. This study evaluates classification accuracy and confusion matrices across multiple SNR conditions. It is found that, at low SNR, and in many of the cases, the hybrid, STFT and wavelet are outperforming the time-based and FFT-based classifier, while all of them perform well at high SNRs. At an SNR of −5 dB, the accuracy performances are 78%, 62.1%, 84.35%, 82% and 90.1% for the time-domain, FFT, STFT, wavelet and the hybrid CNN classifier, respectively, for the test samples considered in this study. The hybrid deep learning that is based on the fusion of the four (time domain, FFT, STFT and wavelet) is always the best in terms of performance. When there is random amplitude, frequency and phase variations in the waveforms, the corresponding accuracy performances at SNR = −5 dB become 68.15%, 49.05%, 78.35%, 80.85% and 89.75% respectively, showing a decrease inaccuracy, but with hybrid still performing well exploiting the advantages from all representations. The results of this experiment are intended to explore the time–frequency representation-level contributions in deep-learning-based waveform classification rather than attributing performance improvements solely to increasingly complex network architectures.
- Open Access
- 24 June 2026
- Published by MDPI AG in Algorithms
- Vol. 19 (7) , 505
Comparative Evaluation of MLP, 1D-CNN and LSTM for Waveform Classification in Additive White Gaussian Noise
Accurate waveform classification in noisy environments is an important task in modern communications, radar signal analysis, biomedical signal interpretation, industrial monitoring and other signal processing systems. This paper investigates the performance of three neural network architectures: Multilayer Perceptron (MLP), one-dimensional Convolutional Neural Network (1D-CNN), and Long Short-Term Memory (LSTM) for multiclass waveform classification in the presence of Additive White Gaussian Noise (AWGN). A time series dataset consisting of multiple waveform classes is generated and corrupted with AWGN across a wide range of signal-to-noise ratio (SNR) levels to simulate noisy signal distortion conditions. The three models are trained and evaluated under identical conditions to ensure a fair comparison. Their classification performance is evaluated in terms of accuracy, Confusion Matrix (CM), Receiver Operating Characteristic (ROC) curve and the Area Under the ROC curve (AUC) across varying SNR values. Simulation results demonstrate that the 1D-CNN effectively captures local temporal patterns and achieves superior robustness in classification at moderate and high SNR levels. The LSTM model demonstrates the ability to capture temporal dependencies in sequential waveform data but exhibits sensitivity to waveform variations due to amplitude, phase and frequency changes and noise at lower SNR values. The MLP, although computationally simpler, shows comparatively limited performance in low-SNR conditions due to its lack of temporal feature extraction capability. For the case of multiclass deterministic waveforms, the accuracy of classification for the 1D-CNN and LSTM is nearly 100% at SNR = 5 dB showing their robustness in classification, whereas the accuracy of MLP is approximately 70% that shows poor classification in noisy conditions. When there is random amplitude, frequency and phase variations in the waveforms, the accuracy of the 1D-CNN and MLP increases with SNR, and 1D-CNN superior to MLP. However, the LSTM accuracy fails to improve with SNR, resulting in poor classification performance in such a scenario. The results provide an insight into the suitability of different neural architectures for waveform classification tasks in noisy communication or other time series applications and highlight the advantages of convolutional feature extraction for robust signal recognition.
- 24 June 2026
- Published by MDPI AG in Algorithms
- Vol. 19 (7) , 505
Open AccessComparative Evaluation of MLP, 1D-CNN and LSTM for Waveform Classification in Additive White Gaussian Noise
Accurate waveform classification in noisy environments is an important task in modern communications, radar signal analysis, biomedical signal interpretation, industrial monitoring and other signal processing systems. This paper investigates the performance of three neural network architectures: Multilayer Perceptron (MLP), one-dimensional Convolutional Neural Network (1D-CNN), and Long Short-Term Memory (LSTM) for multiclass waveform classification in the presence of Additive White Gaussian Noise (AWGN). A time series dataset consisting of multiple waveform classes is generated and corrupted with AWGN across a wide range of signal-to-noise ratio (SNR) levels to simulate noisy signal distortion conditions. The three models are trained and evaluated under identical conditions to ensure a fair comparison. Their classification performance is evaluated in terms of accuracy, Confusion Matrix (CM), Receiver Operating Characteristic (ROC) curve and the Area Under the ROC curve (AUC) across varying SNR values. Simulation results demonstrate that the 1D-CNN effectively captures local temporal patterns and achieves superior robustness in classification at moderate and high SNR levels. The LSTM model demonstrates the ability to capture temporal dependencies in sequential waveform data but exhibits sensitivity to waveform variations due to amplitude, phase and frequency changes and noise at lower SNR values. The MLP, although computationally simpler, shows comparatively limited performance in low-SNR conditions due to its lack of temporal feature extraction capability. For the case of multiclass deterministic waveforms, the accuracy of classification for the 1D-CNN and LSTM is nearly 100% at SNR = 5 dB showing their robustness in classification, whereas the accuracy of MLP is approximately 70% that shows poor classification in noisy conditions. When there is random amplitude, frequency and phase variations in the waveforms, the accuracy of the 1D-CNN and MLP increases with SNR, and 1D-CNN superior to MLP. However, the LSTM accuracy fails to improve with SNR, resulting in poor classification performance in such a scenario. The results provide an insight into the suitability of different neural architectures for waveform classification tasks in noisy communication or other time series applications and highlight the advantages of convolutional feature extraction for robust signal recognition.
- Open Access
- 8 December 2025
- Published by Engineering, Technology & Applied Science Research in Engineering, Technology & Applied Science Research
- Vol. 15 (6) , 30296-30303
Deep Learning-Based Signal Classification in Wireless Fading Channels
Signal detection and identification are critical in today's shared wireless communication environments with diverse devices. These processes are fundamental to wireless, satellite, cognitive radio, and military communication systems. Automatic and real-time identification of radio traffic enables efficient utilization of resources, such as transmission bandwidth and power. It also facilitates dynamic spectrum access by identifying unused spectrum bands and maintaining fair coexistence among the communication devices. Furthermore, it contributes to spectrum surveillance by detecting unknown or unauthorized transmissions, and, hence, enhancing the Quality of Service (QoS) and network performance through effective noise and interference detection and mitigation. In this work, a Deep Learning (DL) approach is proposed for feature extraction and classification of received signals. In particular, a Convolutional Neural Network (CNN) is employed to classify multiple modulation schemes in wireless fading channel conditions. Simulation results show that the CNN provides high classification accuracy, indicating the effectiveness of the proposed method in realistic wireless channels.
- 8 December 2025
- Published by Engineering, Technology & Applied Science Research in Engineering, Technology & Applied Science Research
- Vol. 15 (6) , 30296-30303
Open AccessDeep Learning-Based Signal Classification in Wireless Fading Channels
Signal detection and identification are critical in today's shared wireless communication environments with diverse devices. These processes are fundamental to wireless, satellite, cognitive radio, and military communication systems. Automatic and real-time identification of radio traffic enables efficient utilization of resources, such as transmission bandwidth and power. It also facilitates dynamic spectrum access by identifying unused spectrum bands and maintaining fair coexistence among the communication devices. Furthermore, it contributes to spectrum surveillance by detecting unknown or unauthorized transmissions, and, hence, enhancing the Quality of Service (QoS) and network performance through effective noise and interference detection and mitigation. In this work, a Deep Learning (DL) approach is proposed for feature extraction and classification of received signals. In particular, a Convolutional Neural Network (CNN) is employed to classify multiple modulation schemes in wireless fading channel conditions. Simulation results show that the CNN provides high classification accuracy, indicating the effectiveness of the proposed method in realistic wireless channels.
- 13 August 2024
- Published by Springer Nature
- p. 133-146
LSTM Based Time Series Forecasting of Noisy Signals
- 13 August 2024
- Published by Springer Nature
- p. 133-146
LSTM Based Time Series Forecasting of Noisy Signals
- Open Access
- 1 May 2022
- Published by IJETAE Publication House in International Journal of Emerging Technology and Advanced Engineering
- Vol. 12 (5) , 8-14
A GSM and Arduino Based Fire Detection and Control System
Fire alarm and detection systems play a significant role in preventing life and property damages from fire disasters. Many of the existing fire alarm systems could incur significant expense that is not moderate for low-income users. The main purpose of our project is to design and implement a fire detection and control system that can be produced at a low cost at effective and competitive usage with a main objective of saving life and property damages. The system will be able to detect the temperature of the environment, the presence of fire flame, the gas level and send SMS alert to a user through an inbuilt GSM number. The remote user responds to the received fire STATUS information by turning ON/OFF a fire extinguisher or protection equipment installed at the fire site. The system gets power from a 12 V control supply utilizing a step-down transformer; the Arduino Uno Microcontroller is the core part of the system and it is interfaced with the GSM SIM900 module, flame, temperature and smoke sensors and the fire extinguisher equipment. Keywords—Fire Alarm System, Fire-Detector, GSM Module, Arduino UNO Microcontroller, Temperature sensor, Flame sensor, Smoke sensor.
- 1 May 2022
- Published by IJETAE Publication House in International Journal of Emerging Technology and Advanced Engineering
- Vol. 12 (5) , 8-14
Open AccessA GSM and Arduino Based Fire Detection and Control System
Fire alarm and detection systems play a significant role in preventing life and property damages from fire disasters. Many of the existing fire alarm systems could incur significant expense that is not moderate for low-income users. The main purpose of our project is to design and implement a fire detection and control system that can be produced at a low cost at effective and competitive usage with a main objective of saving life and property damages. The system will be able to detect the temperature of the environment, the presence of fire flame, the gas level and send SMS alert to a user through an inbuilt GSM number. The remote user responds to the received fire STATUS information by turning ON/OFF a fire extinguisher or protection equipment installed at the fire site. The system gets power from a 12 V control supply utilizing a step-down transformer; the Arduino Uno Microcontroller is the core part of the system and it is interfaced with the GSM SIM900 module, flame, temperature and smoke sensors and the fire extinguisher equipment. Keywords—Fire Alarm System, Fire-Detector, GSM Module, Arduino UNO Microcontroller, Temperature sensor, Flame sensor, Smoke sensor.
- Open Access
- 1 August 2020
- Published by ASTES Journal in Advances in Science, Technology and Engineering Systems Journal
- Vol. 5 (4) , 499-504
Remote Control of Garden Plantation Water Pumps using Arduino and GSM Mobile
- 1 August 2020
- Published by ASTES Journal in Advances in Science, Technology and Engineering Systems Journal
- Vol. 5 (4) , 499-504
Open AccessRemote Control of Garden Plantation Water Pumps using Arduino and GSM Mobile
- Open Access
- 1 April 2019
- Published by Institute of Advanced Engineering and Science in International Journal of Electrical and Computer Engineering (IJECE)
- Vol. 9 (2) , 967-972
Implementation of sequential design based water level monitoring and controlling system
In currently decades, water usage either for domestic or commercial purposes is facing critical conditions which negatively affects the sustainability of our environment. From this point of view, reducing water consumptions will participate in the solution of water saving. Monitoring and controlling the tank water level are important functions which effect and work on reduction of water consumption. Many studies in literature focus on proposing a high performance water level monitoring and controlling systems based on either computerized units or microcontroller units. However these systems are characterized by complexity and high cost. In this paper, a highly flexible implemented electronic system is introduced based on sequential design by using general purpose discrete components. The proposed system present simple design of water level sensing stage, in addition, this paper showing full design, simulation, and implementation steps of an automation system which is able to control switching the state of the water pump in accordance to the current level of water in the tank.
- 1 April 2019
- Published by Institute of Advanced Engineering and Science in International Journal of Electrical and Computer Engineering (IJECE)
- Vol. 9 (2) , 967-972
Open AccessImplementation of sequential design based water level monitoring and controlling system
In currently decades, water usage either for domestic or commercial purposes is facing critical conditions which negatively affects the sustainability of our environment. From this point of view, reducing water consumptions will participate in the solution of water saving. Monitoring and controlling the tank water level are important functions which effect and work on reduction of water consumption. Many studies in literature focus on proposing a high performance water level monitoring and controlling systems based on either computerized units or microcontroller units. However these systems are characterized by complexity and high cost. In this paper, a highly flexible implemented electronic system is introduced based on sequential design by using general purpose discrete components. The proposed system present simple design of water level sensing stage, in addition, this paper showing full design, simulation, and implementation steps of an automation system which is able to control switching the state of the water pump in accordance to the current level of water in the tank.
- 1 December 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Automatic water level sensor and controller system
In this paper, we investigate the design of a water level sensor device that is able to detect and control the level of water in a certain water tank or a similar water storage system. The system firstly senses the amount of water available in the tank by the level detector part and then adjusts the state of the water pump in accordance to the water level information. This electronic design achieves automation through sequential logic implemented using a flip flop. A seven segment display and a relay-based motor pump driving circuit are part of this integrated design. The water pump automatically turns on and starts filling the tank when the water level is empty or level ONE and turned-off and stop filling the tank when water level reaches maximum-level NINE; furthermore, the water pump will remain in its standstill state from level EIGHT down to TWO when the level is decreasing due to water consumption.
- 1 December 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Automatic water level sensor and controller system
In this paper, we investigate the design of a water level sensor device that is able to detect and control the level of water in a certain water tank or a similar water storage system. The system firstly senses the amount of water available in the tank by the level detector part and then adjusts the state of the water pump in accordance to the water level information. This electronic design achieves automation through sequential logic implemented using a flip flop. A seven segment display and a relay-based motor pump driving circuit are part of this integrated design. The water pump automatically turns on and starts filling the tank when the water level is empty or level ONE and turned-off and stop filling the tank when water level reaches maximum-level NINE; furthermore, the water pump will remain in its standstill state from level EIGHT down to TWO when the level is decreasing due to water consumption.
- 1 September 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Automatic control of agricultural pumps based on soil moisture sensing
In this paper, we investigate the design and simulation of an electronic system for automatic controlling of water pumps that are used for agricultural fields or plant watering based on the level of soil moisture sensing. The detected signal from the soil moisture sensor is processed by a conditional comparator circuit corresponding to different levels of actual soil moisture content. A logic circuit follows the conditional circuit with its output signals used to activate a system of relays that control the power circuit of the motors used for water pumping. The speed of the motor is varied according to the level of the soil moisture content; the motor is OFF during maximum wet and is running with HIGH speed during dry soil conditions respectively. The duration of water pumping is controlled by a timer circuit where the timer can be designed according to the desired watering time. The different stages of the overall electronic system are simulated and tested using NI MULTISM simulation software. The technique is useful to save human-power, enhances crop or plant productivity, saves water and provides proper water management to ensure efficient and viable farming industry especially in those places that receive less rain or mostly dry areas.
- 1 September 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Automatic control of agricultural pumps based on soil moisture sensing
In this paper, we investigate the design and simulation of an electronic system for automatic controlling of water pumps that are used for agricultural fields or plant watering based on the level of soil moisture sensing. The detected signal from the soil moisture sensor is processed by a conditional comparator circuit corresponding to different levels of actual soil moisture content. A logic circuit follows the conditional circuit with its output signals used to activate a system of relays that control the power circuit of the motors used for water pumping. The speed of the motor is varied according to the level of the soil moisture content; the motor is OFF during maximum wet and is running with HIGH speed during dry soil conditions respectively. The duration of water pumping is controlled by a timer circuit where the timer can be designed according to the desired watering time. The different stages of the overall electronic system are simulated and tested using NI MULTISM simulation software. The technique is useful to save human-power, enhances crop or plant productivity, saves water and provides proper water management to ensure efficient and viable farming industry especially in those places that receive less rain or mostly dry areas.
- Open Access
- 5 August 2015
- Published by North American Publication Corp. in Research Journal of Applied Sciences, Engineering and Technology
- Vol. 10 (10) , 1206-1215
Remote Controlling of Light Intensity Using Phone Devices
In this study, we investigate the design and simulation of an electronic system for remote controlling the level of light intensity of an electrical lighting system used for a certain desired application. The remote controlling is done using the keypad of any Dual Tone Multi Frequency (DTMF) based telephone handset. A DTMF tone command sent from a transmitting fixed or mobile phone terminal will be used to activate a system of relays that control the required level of light intensity. An appropriate active bandpass filters are designed for recognition of the DTMF tones of the transmitted digit. The bandpass filters and the additional subsequent stages of the overall electronic system are simulated and tested using MULTISM. Light intensity controlling is useful to avoid unnecessary electrical power wastage and the remote controlling eliminates the physical presence of a person for adjusting the required level of lighting at the desired location.
- 5 August 2015
- Published by North American Publication Corp. in Research Journal of Applied Sciences, Engineering and Technology
- Vol. 10 (10) , 1206-1215
Open AccessRemote Controlling of Light Intensity Using Phone Devices
In this study, we investigate the design and simulation of an electronic system for remote controlling the level of light intensity of an electrical lighting system used for a certain desired application. The remote controlling is done using the keypad of any Dual Tone Multi Frequency (DTMF) based telephone handset. A DTMF tone command sent from a transmitting fixed or mobile phone terminal will be used to activate a system of relays that control the required level of light intensity. An appropriate active bandpass filters are designed for recognition of the DTMF tones of the transmitted digit. The bandpass filters and the additional subsequent stages of the overall electronic system are simulated and tested using MULTISM. Light intensity controlling is useful to avoid unnecessary electrical power wastage and the remote controlling eliminates the physical presence of a person for adjusting the required level of lighting at the desired location.