2018-2019

Project Title: Design of a Wearable EEG Device For Diagnosing Neurological Disorders Using Artificial Intelligence
Students: Esther Tabiri Owusu-Ansah & Richard Adekponya Kabu
Supervisor: Dr. Robert A. Sowah

Abstract:
In many low-income and middle-income countries such as Ghana, there is a huge gap between availability of mental health care and delivery of said service due to the insufficient number of neurologists and other certified mental health personnel. This impedes the diagnosis and treatment of many neurological disorders.
This project aims to develop a system to help in the monitoring and diagnosis of patients with neurological and mental disorders by developing a wearable EEG device for the acquisition of brain signals and a deep neural network for the classification of signals obtained.
The hardware device comprised of a power supply unit, EEG sensor electrodes for detecting and recording brain signals, pre-filter board for amplifying and filtering the signal acquired, DSP processor to convert the signal from analog into digital form and a microcontroller for signal transmission and processing. A remote server developed in Django was used for integration with the EEG device. MNE and PyEEG were used for the signal processing and feature extraction in Python. The neural network was developed using Tensorflow. Mobile and web applications were built using React-Native and Laravel to enable health experts and patients interface with the system developed.
After integration of both the hardware and software modules, 60 second EEG recordings were obtained from the electrode cap at the A1, A2, FP1, FP2, F7 and F8 scalp locations. These signals were then transmitted to the Django server where they were successfully saved to the database. The signals were processed and 10 features were extracted for classification. The signals were classified with an accuracy of 88%. The classification results and EEG signals were retrieved from the database by the web and mobile applications developed and adequately visualized in both the patient and doctor views.
In conclusion, a complete and efficient patient monitoring and diagnostic system was developed, which would go a long way to save lives and ensure healthcare needs are met. Classification using other features may be tested and the number of electrode could be increased in future works in order to increase the accuracy of the prediction model.

Project Title: Secure Wireless Home Automation With OpenHAB (Open Home Automation Bus) 2
Students: Rexford Addo,  Dale Eyram Boahene & Owoh Dalton Chukwuezugo
Supervisor: Dr. Robert A. Sowah

Abstract:
OpenHAB 2, the latest version of OpenHAB software, runs at the center of smart homes. Its benefits to homes include effective and efficient controls, safety, convenience, comfort, and peace of mind. These benefits are limited by key factors such as platform fragmentation, lack of technical standards, proprietary protocols, and failure of vendors to import old devices with patches and updates.

OpenHAB 2 is key to sustainable entry into the Smart Home ecosystem, thus, posing material questions to scholars, investors, households, and the government of Ghana. This research project work seeks to design, develop and deploy a protype for Secure Wireless Home Automation System with OpenHAB 2. We employed the use of two (2) high performance microcontrollers; the Arduino Mega 2560 and Raspberry Pi Model B to develop a prototype of an automated smart home. The Arduino microcontroller interfaced with a 16-channel relay acted as the switching module, while the Raspberry Pi microcontroller running on the OpenHAB software functioned as the server. In designing a wireless controlled switch for home appliances, two security procedures were implemented namely: the token-based and the JWT authentication procedures.

Further evaluation of existing home automation systems showed similarities with the system developed in this research work, addressing identified key gaps such as security features, remote control functionalities, provision of an interactive mobile, including web application that resulted in performance improvements.

Upon testing, our Android Application performed well with no compatibility noise among the popular mobile Operating Systems. It exhibited high battery life savings of about 10% over the average data-enabled application, passed all remote-control functionalities from both web and mobile apps. The web service was tested to meet basic data validation and response data requirements needs which was fully compliant. Security testing on the OpenHAB framework was demonstrated as an independent deployment before integration into OpenHAB. In the project prototype, the JWT authentication provided robust security and waded off intrusion of the developed smart home automation system. This was evidenced in the trials demonstrated yielding 100% effectiveness rate.

This work recommends among other things, that old devices in the home automation ecosystem should be made less vulnerable by mandating vendors to patch old devices provide periodic updates. Availability of hard and soft infrastructure is key. The Government should undertake a pilot roll out of homes with secured wireless automation with OpenHAB 2. Appropriate policy and legal framework to enable adoption of home automation technologies should be put in place by the Government of Ghana. Nonetheless, future work may explore how OpenHAB 2 could be deployed to assist the automation of urban transportation in Ghana.

Project Title: Development of a Mobile Application for the Detection and Categorization of Arrhythmia Using ECG Signals
Students: Danielle Naa Djaa Mills & Israel Nanor
Supervisor: Dr. Godfrey A. Mills
Abstract:

Arrhythmia is a condition of the heart where the heart rate or rhythm of a patient is abnormal. During an arrhythmia, the heart can beat too fast, too slowly or erratically. An arrhythmia like Ventricular fibrillation, which is an erratic, disorganized firing of impulses from the ventricles is a leading cause of most sudden cardiac deaths. Early and timely detection of arrhythmic events may lead to avoidance of any catastrophe or loss of human life. Besides, if it detected early, remedial action could be taken for treatment where necessary.

This project provides a mobile based solution for the detection and categorization of arrhythmia. A modified Pan-Tompkins algorithm was developed for identification and extraction of P and T waves alongside the QRS complexes from the ECG signals. An AI technique based on the Random Forest algorithm was developed to classify 12 different arrhythmias. The classifier was trained with 1300 randomly generated samples.

To test the performance of the arrhythmia detection system, different sets of data from the MIT-BIH arrhythmia database were used. For each dataset the following features; RR interval, QRS duration, PR interval, P wave duration, T wave duration, QT interval, Heart rate as well as the amplitudes of P, Q, R, S and T waves were extracted. Results show that the system was able to adequately categorize the selected conditions of arrhythmia with an accuracy of 100%.

Project Title: Development of an Automatic Switching System for Multiple Power Sources With App Access
Students: Bright Kusi Appiah & Seth Ofori-Amanfo
Supervisor: Dr. Robert A. Sowah

Abstract:
In many developing countries like Ghana, electric power generated by the utility supply authority is inadequate to meet the demands of the people and, it is sporadic. The need for electricity is increasing day by day and the frequent power cuts are causing many problems in different areas like banks, colleges/schools, hospitals, houses and industries.  To solve this problem, manual changeover systems have been deployed. These systems are challenged in that the manual changeover does not guarantee a continuous power supply in the case of power cuts.
The main aim of this work is to tackle this problem by designing and implementing an automatic switching system between the mains and a solar source of power supply in order to provide an uninterrupted power supply to connected loads. This system has an app access incorporated so as to give its users the flexibility of selecting their preferred source of power supply.
The system comprises of an Arduino UNO microcontroller and two Solid State Relays (SSR), which together controls switching between the two connected sources of power supply. The system also has voltage sensors which inform the Arduino microcontroller about the voltages coming from each source; with which switching operations are done. The Bluetooth module connected to the system helps the system interface with the app designed for manual overrides of the system by users.
The system prototype was implemented and tested with two sources of power supply with a threshold voltage of 210V, upon which switching decisions were made. At system start, the grid was the primary source of power supply with a voltage reading of 230V. Upon the system’s detection of a voltage drop below 210V, the Solar source of power supply was switched to, within two seconds. Whenever the mains’ voltage was above 210V, the system reverted back to it. Manual overrides done with the mobile app took effect almost immediately.
The system worked effectively with a 90% success rate with a guarantee that power supply to connected loads are not interfered with. The system is well suited for homes, work places and hospitals.

Project Title: Design of a Wearable EEG Device For Diagnosing Neurological Disorders Using Artificial Intelligence
Students: Conversion of Conventional Lathe Machine to Computer Numerically Controlled (CNC) Machine
Supervisor: Dr. Robert A. Sowah

Abstract:
Computer Numerically Controlled (CNC) machines play an integral and irreplaceable role in manufacturing industries and it does not only realize rapid industrial production, but also saves manpower and material resources. The conventional machining tools such as the manually operated lathe machine is not very efficient and is quite challenging when used for commercial purposes due to low production capability, hence the need for CNC machines for enhanced and productivity in industrial automation.
Currently, there are many conventional lathe machines being used in industries in the country. The need to migrate the manual operations of these machines to full or semi-automation status is of utmost importance. It is required to convert these machines to semi-automatic lathe by retrofitting, in order to grow the country’s development in the machining industry. Converting the manual lathe into a semi-automatic lathe required three significant portions to be modified namely electronics, mechanics and the hydraulic components.
In this project, we propose a computer numerically controlled automation systems for the conversion of the conventional lathe machine by developing both mobile and web applications to perform the automation tasks. The actual conversion process was first carried out on a mini center lathe machine at the Ghana Atomic Energy Commission at Kwabenya in Accra. Firstly, unnecessary components such as the gearbox and handwheels were removed and mounts for the various motors for the automation tasks were added to the machine as part of the mechanical system component modification. The electronic component modification involved setting up of stepper motors for the X, Y and Z axes of the lathe machine. The Arduino microcontroller is used to control the motors via the motor drive systems. At the end of the conversion, the newly retrofitted machine was able to machine a designed part successfully when the machining instruction in G-code format was sent to the microcontroller, the instruction was decoded and executed successfully.