Laryngectomised people (LP) have many problems with speaking in terms of intelligibility, volume and effort it takes. This thesis describes the design and simulation of a voice enhancement device which will improve the speech of LP with respect to intelligibility and volume. It only describes the chosen hardware peripherals in order to record, process, amplify, and output the modified speech. The recorded audio could be processed using a digital signal processor (DSP) that could be programmed such that it will filter the undesired noise and shift the pitch of the speaker's voice in real-time. This thesis focuses on the real-time pitch shifting of the voice, the hardware choices and implementation of the sound output of the system, and the implementation of DSP of the Tracheo-Esophageal Speech Signal Amplifier (TESSA). The real-time pitch shifting makes use of the Ocean algorithm, of which the principle is described, implemented and tested. Intelligibility tests are performed in order to find the optimal shifting parameters.

Many countries in the world face the problem of fruit frost during spring, which results in the loss of harvested crops. Solutions exist to prevent the fruit to freeze, however this requires that the temperature is known. Therefore, a smart autonomous temperature sensor network is designed that is capable of acquiring a 3D temperature profile of a fruit orchard to detect fruit frost. This thesis discusses the design of the energy harvesting and control module, which is one of the three sub-modules of the system. The goal is to sustain the energy demands of the sensor in a durable way, in order to operate for 20 years without requiring maintenance. To achieve this, several ambient energy sources and their harvesting techniques are investigated and simulated to examine the energy that can be harvested. This resulted in the use of solar energy, harvested by a solar cell and controlled by a maximum power point tracking module. Next, different energy storage implementations are considered to store the harvested energy temporarily. The use of a 5 F supercapacitor is concluded. Furthermore, an energy monitoring circuit is designed to measure the energy stored in the supercapacitor. Full system reliability simulations are done to verify the complete design by utilising the solar irradiance and temperature data of the past ten years. From these simulations is concluded that the design can sustain the energy demands of the sensor at all times.

In this thesis, the design and implementation of a wireless communication module and an accompanying network are discussed. This wireless communication module is used in a device which communicates frost temperatures measured at orchards. The goal is to gather data on how fruit frost develops on orchards, and to warn a fruit farmer if it starts freezing. First, several wireless communication protocols are discussed and the one best suited for the application is chosen. It is decided to use the LoRaWAN communication protocol. Then, the off-the-shelf hardware components which are required to implement the LoRaWAN communication protocol are chosen. Furthermore, LoRa communication parameters and the LoRaWAN network structure are discussed. A scheduling system is designed and proposed to increase the reliability of the network. At last, network simulations are performed to verify the chosen implementation.

One of the main problems with Instance-level Image Retrieval in video data is that for longer query videos or large amount of image queries, comparing all of the query images to every extracted frame is time-inefficient. This thesis aims to solve this problem by implementing Nearest Neighbour Search (NNS) algorithms and data compression methods, significantly reducing total comparison time. In most NNS use cases, the reference data is provided before reaching the user, allowing methods such as ANNOY or HNSW to partition the data beforehand. However, little research has been done into partitioning the data during run-time. In this thesis, the use of Nearest Neighbor Search and Data Compression methods are discussed for the purposes of matching a query image to a query video, both of which are provided at run-time. The result is an implementation of several state-of-the-art NNS and data compression methods in a system which, based on the amount of query images and the amount of extracted keyframes, selects the optimal comparison method to be used, as well as its optimal parameters if applicable.

This report details the design and implementation of a subsystem providing a web-based platform for smartwatch data acquisition. A smartwatch application is developed to read out the accelerometer, gyroscope and heart rate sensor on the smartwatch and transmit the sensor data to the platform. A platform is developed to receive the data and store it in a database. Recordings can be downloaded from the platform to use in research of human activity. The platform presents the type of activity the smartwatch user is doing using machine learning models, by integrating the other subsystem. Multiple tests have been performed to analyse and improve the performance of the system. The battery life of the smartwatch has been tested using various settings in the smartwatch application to determine the most power-efficient settings.

Fruit-frost during spring is one of the main causes of damage to the harvest of fruit in orchards. Various systems using different methods of preventing spring-frost are available on the market. To determine when these systems should be activated, the temperature in the orchard needs to be determined. In this project, a self-sustaining autonomous temperature sensor network is designed, which is capable of making a 3D temperature map of the orchard. The system is used to warn farmers when the threat of fruit-frost occurs and to gather data on the spatial variation of the temperature in the orchard. In this thesis, the focus is put on designing the smart measurement and control system. This includes choosing an appropriate control unit and temperature sensors. Also, the software for the control unit is designed, which allows for a smart measurement scheme that balances energy usage and measurement frequency. Finally, an estimation of the energy usage of the subsystem is given based on theoretical analysis.

This report entails one of two subsystems in a joint project to provide a web-based platform for smartwatch data acquisition, for applications in healthcare. In this work, we design and implement algorithms for human activity recognition using various machine learning approaches and test them on data acquired online as well as using our own developed platform. Together with the web-based platform, this provides a solid base for more research using data gathered from smartwatches. The human activity recognition is implemented first using a classical machine learning approach with feature extraction and a random forest classifier. Next, both convolutional neural network and a recurrent neural network are implemented using Tensorflow [1]. We further perform several tests to investigate: (i) the optimal segment size with respect to classification accuracy, (ii) the effect of filtering and preprocessing on the classification results, and (iii) the best classifier for activity detection.