Search

This paper presents drone classification at millimeter-wave (mmWave) radars using the deep learning (DL) technique. The adoption mmWave technology in radar systems enables better resolution and aid in detecting smaller drones. Using radar cross-section (RCS) signature enables us to detect malicious drones and suitable action can be taken by respective authorities. Existing drone classification converts the RCS signature into images and then performs drone classification using a convolution neural network (CNN). Converting every signature into an image induces additional computation overhead; further CNN model is trained considering fixed learning rate. Thus, when using CNN-based drone classification under a highly dynamic environment exhibit poor classification accuracy. This paper present an improved long short-term memory (LSTM) by introducing a weight optimization model that can reduce computation overhead by not allowing the gradient to not flow through hidden states of the LSTM model. Further, present adaptive learning rate optimizing (ALRO) model for training the LSTM model. Experiment outcome shows LSTM-ALRO achieves much better drone detection accuracies of 99.88% when compared with the existing CNN-based drone classification model.

The modern world does not stand still. We used to be surprised that technology could speak, but now voice assistants have become real family members. They do not simply turn on the alarm clock or play music. They communicate with children, help solving problems, and sometimes even take offense. Since all voice assistants have artificial intelligence, when communicating with the user, they take into account the change in their location, time of day and days of the week, search query history, previous orders in the online store, etc. However, voice assistants, which are part of modern smartphones or smart speakers, pose a threat to their owner’s personal data since their main function is to capture audio commands from the user. Generally, AI smart speakers such as Siri, Google Assistance, Google Home, and so on are moderately harmless. As voice assistants become versatile, like any other product, they can be used for the most nefarious purposes. There are many common attacks that people with bad intentions can use to hack our voice assistant. We show in our experience that a laser beam can control Google Assistance, smart speakers, and Siri. The attacker does not need to make physical contact with the victim’s equipment or interact with the victim; since the attacker’s laser can hit the smart speaker, it can send commands. In our experiments, we achieve a successful attack that allows us to transmit invisible commands by aiming lasers up to 87 m into the microphone. We have discovered the possibility of attacking Android and Siri devices using the built-in voice assistant module through the charging port.

Personalized diagnosis of chronic disease requires capturing the continual pattern across the biological sequence. This repeating pattern in medical science is called "Motif". Motifs are the short, recurring patterns of biological sequences that are supposed signify some health disorder. They identify the binding sites for transcription factors that modulate and synchronize the gene expression. These motifs are important for the analysis and interpretation of various health issues like human disease, gene function, drug design, patient's conditions, etc. Searching for these patterns is an important step in unraveling the mechanisms of gene expression properly diagnose and treat chronic disease. Thus, motif identification has a vital role in healthcare studies and attracts many researchers. Numerous approaches have been characterized for the motif discovery process. This article attempts to review and analyze fifty-four of the most frequently found motif discovery processes/algorithms from different approaches and summarizes the discussion with their strengths and weaknesses.

Clustering is an energy-efficient routing algorithm in a sensor cloud environment (SCE). The clustering sensor nodes communicate with the base station via a cluster head (CH), which can be selected based on the remaining energy, the base station distance, or the distance from the neighboring nodes. If the CH is selected based on the remaining energy and the base station is far away from the cluster head, then it is not an energy-efficient selection technique. The same applies to other criteria. For CH selection, a single criterion is not sufficient. Moreover, the traditional clustering algorithm head nodes keep changing in every round. Therefore, the traditional algorithm energy consumption is less, and nodes die faster. In this paper, the fuzzy multi-criteria decision-making (F-MCDM) technique is used for CH selection and a threshold value is fixed for the CH selection. The fuzzy analytical hierarchy process (AHP) and the fuzzy analytical network process (ANP) are used for CH selection. The performance evaluation results exhibit a 5% improvement compared to the fuzzy AHP clustering method and 10% improvement compared to the traditional method in terms of stability, energy consumption, throughput, and control overhead.

An ad hoc network is a wireless mobile communication network composed of a group of mobile nodes with wireless transceivers. It does not rely on preset infrastructure and is established temporarily. The mobile nodes of the network use their own wireless transceivers to exchange information; when the information is not within the communication range, other intermediate nodes can be used to relay to achieve communication. They can be widely used in environments that cannot be supported by wired networks or which require communication temporarily, such as military applications, sensor networks, rescue and disaster relief, and emergency response. In MANET, each node acts as a host and as a router, and the nodes are linked through wireless channels in the network. One of the scenarios of MANET is VANET; VANET is supported by several types of fixed infrastructure. Due to its limitations, this infrastructure can support some VANET services and provide fixed network access. FANET is a subset of VANET. SANET is one of the common types of ad hoc networks. This paper could serve as a guide and reference so that readers have a comprehensive and general understanding of wireless ad hoc networks and their routing protocols at a macro level with a lot of good, related papers for reference. However, this is the first paper that discusses the popular types of ad hoc networks along with comparisons and simulation tools for Ad Hoc Networks.

Aiming at the defects of low speed and high consumption of traditional wireless sensor and actuator networks, this paper studies the wireless sensor and actuator network repair algorithm based on graph theory. This paper first analyzes the concept of the wireless sensor and sensor network, and then presents the cooperation algorithm in the wireless sensor and sensor network. The NS2 simulation model topology and truncation points are basically fixed. The simulation results show that the real-time performance and power consumption of the proposed algorithm are better than comparing the K-IDS algorithm and the RCR algorithm.

The automatic identification system is used by the ships, and the VHF data link (VDL) mode 4 system is used by the avionics industry. Based on ground infrastructure, the traditional surveillance applications for ships are built with radar support. The major drawback of radar is the inability to see behind large obstacles. This obstacle affects the attenuation signal at receiving ship. Therefore, we present an obstacle-based radio propagation model that considers the effect caused by the presence of obstructing ships in line of sight. This model is evaluated in terms of throughput, collision, and packet transmission under distributed slotted MAC.

Employing the growing IoT technologies with efficient biometric and motion sensors, the prototype is developed to address one of the most critical issues that humanity faces today. It is not only a locomotive aid that enables the disabled and the elderly to commute with ease, but also a safe way of health monitoring, and a gateway to the digital world of endless possibilities. The plight of people with disabilities and the elderly has continued to evoke serious concern as a disabled life in a different world of helplessness and deprivation. While many hi-tech products available in the market aim at addressing different issues of the disabled, due to the sheer cost and inefficiency of such solutions, they have failed to reach the masses. This product is being developed with an intention to provide complete self-dependence to disabled people and the elderly in performing everyday work at extremely low cost. Locomotive assistance is provided by a customizable-accelerometer/voice command/IOT/touchscreen-enabled wheelchair, which can either be controlled by the disabled person himself, or by remote assistance (using IoT), supported by live camera feed. The personal assistant robot shall help in performing tasks like cleaning & mopping, moving objects like medicines, parcels, food, water, etc. This robot can be controlled by Voice control, RF module, or by Bluetooth interface (customizable) and shall optionally provide live camera feed too. The digital home aspect of the project shall automate the lighting & cooling. The door access, light, fan, entertainment system, power sockets, and other household appliances can also be controlled wirelessly using an IR remote. The emergency alert system shall work by manual control. This alert system shall notify the support teams in case of an emergency.

Face recognition has been widely studied and studied for many years, but most PC-based face recognition systems have very limited portability and mobility. Face recognition is a process of dynamically capturing facial features through a camera connected to a computer and simultaneously comparing the captured facial features with the facial features previously entered into the personnel library. Face recognition-based person authentication system has been popular among other biometrics recently. This technology can be applied to important departments such as public security, banking, and customs to provide convenient and efficient detection methods. In this paper, we discuss a method of access control system using face recognition technology for entrance limited places. Some face recognition technologies that have shown state-of-the-art performance at their time are discussed. Here we present a method for access control system using facial recognition technology.

Apple defect detection using hyperspectral image has become an interesting research focus since the last decade. It is important for wide range of applications, such as agricultural sector, food processing, and automatic fruits grading system. However, this task is a challenging one due to calyx and stem, their different types and orientation, and similar in their visual appearance. The proposed method based on a deep learning approach using SqueezeNet architecture. However, the apple images are extracted and fed into a deep network for training and testing. The proposed SqueezeNet architecture utilizes convolution neural network to regress a bypass connection between the fire modules across the images. It has been evaluated our own created dataset. The excremental result shows that our proposed methods are efficient and effective. The general detection rate was 92.23%.

Ansel Adams once said, “When words become unclear, I shall focus on the photographs.” The images hold more power of expression than words, and thus human face conveys more meaning with expression. Face expressions play an inevitable role in verbal as well as non-verbal communication. Based on same philosophy, the IT industry is also showing keen interest in developing the technology to better understand the human expressions or human emotions. Big IT giants, like Google Inc., are spending a huge portion of their budget and manpower to develop the face expression recognition (FER) systems. However, the idea of developing this area for behavioral studies and in medical rehabilitation is still a challenge. Since every face holds a different story, studying the same using existing technology and understanding toward machine learning is a vast area to develop. Thus, standardizing the facial expression understanding for the computers becomes the major issue in the current environment. ‘AI-Based Mood Enhancer—An Android App’ is an attempt to make use of machine learning and virtual assistant technology, and making the computer systems ready for better understandings on facial expressions. This would be one of its kind applications that will communicate with its users, based on the user’s facial expressions. The application, which will run on the Android mobile devices, will capture the video stream of the user’s facial expressions using the device front camera. The inputs will then be processed using the machine learning algorithms of the open-source libraries like Weka, OpenCV, Dlib, and TensorFlow Lite. The unique feature of this application is that, based on the evaluated expression classification, the application will attempt to share the dynamic response to better and enhance the user’s mood, hence named as ‘The Mood Enhancer.’

Fruit detection and classification are a challenging task in image processing. This paper presents a novel approach to fruit detection using deep convolutional neural networks. The aim is to build an accurate, fast, and reliable fruit detection system, which is a vital element of an autonomous agricultural artificial intelligence platform; it is a key element for fruits yield detection and automated fruits processing industry. Recent work in deep neural network has led to the development of a state-of-the-art object detection and classification. This paper demonstrates the design and implementation of deep learning-based automated categorization of the apple images captured from fruits processing industry. The system is based on a convolutional neural network (CNN) followed by the selection of proposed regions. The training of the classifier is performed, with a dataset derived from the set of images taken from the agricultural-based industrial fruit sorting process. In the preparation of the CNN architecture model, initializing the parameter configuration accelerates the network training process. The results of the experiments using CNN algorithm showed the performance of defect detection on the apple fruit of 96%.