▼ Fast development of tracking devices has made trajectory outlier detection(TOD) possible and meaningful. Given a set of trajectories T, a TOD algorithm outputs a subset of T, of which trajectories are different from mostof the other trajectories in some aspect(s). These trajectories, namely outliers, can indicate important or interesting information and are thus worthnoticing. TOD techniques can be used for surveillance security, accidentdiscovery, and many other purposes. Many of the existing TOD algorithms consider only spatial trajectoryoutliers. They can detect trajectories that follow an abnormal route ordirection. While some existing algorithms are capable of detecting outliersin temporal aspects, like trajectories with abnormal time duration or speed,they have their own weaknesses. For example, they can be computationallyexpensive, or fail to detect important types of outliers. In this work, we aimto overcome these shortcomings of previous TOD algorithms. A novel grid-based TOD algorithm is proposed that is capable of detectingtemporal-spatial outliers including density, direction, duration, and speedoutliers with accuracy as well as fast calculation. The algorithm performsthe following three main steps: (i) it calculates density, direction, duration,and speed features of all trajectories in the input set T; (ii) it transformsfeature information of trajectories into grid information; (iii) it examineseach trajectory grid cell by grid cell. Following these three steps, outlyingtrajectories are extracted. By conducting experiments on several data setsincluding both simulated and real ones, the algorithm is shown to be efficientin detecting density, direction, duration, and speed outliers. It outperformsstate-of-the-art TOD algorithms in various aspects. Advisors/Committee Members: Langendoen, Koen (mentor), Delft University of Technology (degree granting institution).

▼ 

The use of Light Emitting Diodes (LED) as a main source of illumination can help in significantly reducing global energy conspumtions. The LED is well suited for modulating visible light, enabling a wireless communication channel. A popular topic in Visible Light Communication (VLC) research is to push data throughput past the LED bandwidth limitation of around 4-5 MHz. Less attention has been given to systems using cheaper and slower components. This thesis focuses on effectively utilizing the available bandwidth in low cost dimmable VLC systems, to allow for increased data throughput, where the system bandwidth is limited by the receiver. Non- ickering constant perceived light intensity must furthermore be maintained under variable dimming conditions, to ensure positive user experience. A novel modulation scheme for VLC systems with dimming support is pro- posed. Adaptive Rectified Pulse-Width Modulation (ARPWM) modulates the light source with fine-grained pulses within a defined symbol period. The symbols are demodulated as light intensity levels. This allows for sending multiple bits per symbol in a manner similar to Pulse Amplitude Modula- tion (PAM). Upper and lower boundaries are applied to the duty cycle of the pulses according to the desired dimming level, reducing the number of light compensation symbols needed to maintain a constant perceived light intensity for the user. ARPWM is compared to existing modulation schemes, and a throughput increase by a factor of 4.3 was observed compared to state-of-the-art Vari- able On-Off Keying (VOOK). With the testbed used for the experiments we have demonstrated the use of ARPWM in an indoor office environment, with a maximum communication distance of 4.2 meters in a low-noise en- vironment. ARPWM is shown to offer valuable increase in data troughput at a low complexity level. The relatively high SNR requirements of ARPWM make it best suitable for short-range indoor VLC.

Embedded Systems

Advisors/Committee Members: Langendoen, Koen (mentor), Delft University of Technology (degree granting institution).

▼ 

More and more devices are gaining some form of smartness either by sensing their environment on their own or by communication. An example of this is Smart Lighting, a field in which Chess Wise is active for over 6 years. By adding sensors and communication lighting can be controlled more efficiently. For example motion sensors to turn off lighting where no movement is detected, or a light level sensor to achieve the same light level throughout the day. This saves energy when natural light is available. Extending the wireless network with a gateway to the internet opens even more possibilities to control and monitor the wireless network. The technology used by Chess, called MyMesh, is a proprietary mesh network. One node is a small embedded device with limited resources. The MyMesh network is selforganizing. All nodes have the same function in the network. All nodes cooperate to get messages from node A to node B. Being selforganizing allows nodes to be added or removed at any time. By leveraging the communication between nodes their limited resources can be combined into a big shared resource. The MyMesh network currently uses an protocol, SharedState, to combine the limited memory capacity (cache) of the nodes. During communication nodes exchange memory items, called Tokens. By moving these tokens around each node has access to the combined memory space of the whole network. In this thesis a renewed algorithm is proposed that improves the overall storage capacity of the network. This renewed algorithm also scales better with the number of nodes compared to the current algorithm. SharedState used a probabilistic approach to transmit tokens and replace tokens in the cache. With the renewed algorithm per cache entry two additional fields (age, rx) are added. These fields are stored to calulate the informational value by Shannon[19]. The age field indicates how long ago the token was placed in the cache. The rx indicates how many times the token was received during that time. Using the age, rx and the average number of tokens received per round the informational value can be calculated. This informational value is higher for tokens that are received less often. Each round the cache is reordered saving the tokens with the highest informational value at the top. Tokens that do not fit in the local cache are discarded. These discarded tokens carry a low informational value. To transmit tokens the renewed algorithm uses the cache position as the probability of ending up in the transmit message. The higher the position in the cache, and the higher the informational value, the more likely a token is to end up in the transmit message. Using this new proposed algorithm the storage capacity now grows with the number of nodes. And more tokens are retained in smaller networks.

Electrical Engineer | Embedded Systems

Advisors/Committee Members: Langendoen, Koen (mentor), Delft University of Technology (degree granting institution).

▼ 

A robot swarm of zebros (Dutch: ZEs Benige RObots) is being developed at the TU Delft, with the purpose of forming a large self-deploying sensor network that survey remote locations without the need for any pre-existing infrastructure, or be used for evaluating swarming algorithms in the field. Towards this goal, zebros need to be able to communicate with and localise nearby neighbours. Existing localisation solutions showed to be inadequate for the task, because they generally do not allow for localisation without existing infrastructure, do not allow scaling to a large swarm or because they rely on robot behaviour, which introduces a dependency of localisation on a specific swarming algorithm. We present a fully-localised method of localisation called Tangolation, which estimates the location of a neighbour as a range and an angle, accompanied by a confidence value from 1 (low) to 5 (high). With Tangolation, two nodes estimate each other’s next location from a range measurement using Two-Way Ranging (TWR), and the exchange of the nodes’ displacement since the last range measurement. Therefore, the two nodes only need each other to determine each other’s location as long as at least one of the two is moving, allowing Tangolation to scale well to larger swarms. No knowledge about the network topology or control over robot behaviour is needed, and communication between localising nodes only needs to happen sparsely, at a frequency of once every five seconds per neighbour. A fully distributed Time Division Multiple Acces (TDMA) protocol named Anarchic TDMA (AN-TDMA) was devised to support communcation within the robot swarm, where nodes synchronise the slot start times without the need for a special coordinating node that indicates the start of slots or frames. AN-TDMA showed to increase the reachable channel utilisation ratios by 55% to 67% compared to the ALOHA method recommended by Decawave. For TWR, the Decawave DW1000 transceiver was used, and the BNO055 IMU aided in displacement measuring. Tangolation was tested in a simulation with realistic conservative estimates of the noise over the range and displacement measurements. Even using conservative noise estimates, at least 95% of the estimated location angles are within the required 22◦ of the true value. Depending on the confidence level that Tangolation reports, 65% to 84% of the angle estimates are within 10◦ of the true value, which shows that Tangolation more than meets the set requirements.

Electrical Engineer | Embedded Systems

Advisors/Committee Members: Langendoen, Koen (mentor), Noroozi, Arash (mentor), Verhoeven, Chris (mentor), Delft University of Technology (degree granting institution).

▼ 

The Delfi-n3Xt was the second nanosatellite developed at the Delft University of Technology, and launched in 2013. Its successor Delfi-PQ is expected to be launched in first half of 2019. The Attitude of a satellite can be referred to as its orientation in space with respect to inertial reference frame. The Delfi-n3Xt was the first satellite from Delft University of Technology, to include three-axis Attitude Determination and Control System/Subsystem (ADCS). It was designed with 5 modes of operation. Four of these were advanced modes. In addition, the Delfi-PQ is not intended to include advanced modes of operation. Hence, this thesis considers using the Delfi-n3Xt ADCS software. This software is extended as a baseline implementation on MSP-EXP432E401Y launchpad. Nearly, 32 % of total nominal power is assigned to ADCS. Hence, energy efficient design alternatives could be considered for future satellite missions. In addition, ADCS is a critical subsystem, failure of ADCS means failure of satellite mission. This thesis aims to improve performance and energy consumption of ADCS. This thesis considers study of three different Digital Signal Processing (DSP) alternatives: Double Precision (DP), Single Precision (SP) and Fixed Point (FxP) arithmetic. Study in this thesis concludes that FxP alternative provides approximately 6.7 times better performance, and approximately 7 times better energy efficiency over baseline. Hence, this thesis proposes the use of FxP DSP alternative. It was also concluded that, the SP arithmetic has equivalent accuracy compared with DP. Moreover, SP provides approximately 3 times better performance, and approximately 2.7 times better energy efficiency over baseline. Therfore, future implementations could benefit from an SP alternative. A major part of the ADCS power is allocated to sensor and actuator. This leaves only 10 % of the total nominal power assigned to ADCS software. Hence, the proposed alternative might not provide considerable improvements on total nominal power. However, for future satellite missions, if there exists a computationally intensive algorithm assigned with more significant amount of total nominal power, then, the proposed alternative could serve as an initial study. However, this does not guarantee that the suggested alternative could satisfy more accurate requirements. In such case, FxP implementation might result in accuracy violation. And use of SP alternative is proposed. In such case, a new study is suggested in order to benefit from FxP alternative.

Delfi-PocketQube

Embedded Systems

Advisors/Committee Members: Wong, Stephan (mentor), Fónod, R. (graduation committee), Langendoen, Koen (graduation committee), van Genderen, Arjan (graduation committee), Delft University of Technology (degree granting institution).

▼ 

Tiny batteryless sensing systems that do not have a constant supply of power have become an increasingly appealing way to gather data, in hard to reach places, at low cost. These systems suffer from frequent and unpredictable power loss. Preserving information through power failures is an ancillary challenge. Unfortunately, state-of-the-art systems do not fully tackle this issue. We introduce a system that offloads the implementation burden and evaluate it against state-of-the art alternatives.

Embedded Systems

Advisors/Committee Members: Pawelczak, Przemek (mentor), Langendoen, Koen (graduation committee), Panichella, Annibale (graduation committee), Delft University of Technology (degree granting institution).

▼ 

Automated generation of robot controllers using an Evolutionary Algorithm(EA) has received increasing attention in the last years as it has the potentialfor a reduction in the development time of a robot. Often these EAs generateNeural Networks (NNs) as robot controllers. Using a NN for automaticallygenerating robot controllers has two important downsides: 1.) A human isnot able to fully understand the inner working of a multi-layer NN, and 2.)a NN has only limited abilities to decompose a complex task into sub tasks.Both of these downsides can be addressed by using a State Machine (SM)instead of a NN as robot controller. Therefore, this thesis introduces an EAcalled Evolving State Machines As Controllers (ESMAC). ESMAC generatesSMs instead of NNs. A SM is understandable for humans because ofits modularity and allows for task decomposition by using a state for eachsub task. Furthermore, two extensions of ESMAC are proposed: adaptiveESMAC and selector ESMAC. Adaptive ESMAC aims to automatically determinesthe number of states with which the best tness for a task canbe achieved. Selector ESMAC replaces the transitions that are used in aSM to switch between states with a NN-based switching mechanism. This switching mechanism allows mutations to make more gradual changes to aSM's behaviours, which improves the performance of the EA. The performance of ESMAC is evaluated on two robotic tasks: come-and-go and phototaxis-with-obstacles. All three variants of ESMAC showequally good performance as a NN-based EA on the evaluated tasks. Thecontrollers generated with standard ESMAC and adaptive ESMAC hardlymake any state transitions and mainly use one state. However, controllers that do use multiple states appear to be more robust to changing scenarios and in noisy environments. Selector ESMAC is able to generate SMs-based controllers that have complementing states and, therefore, shows potentialfor decomposing a task into sub tasks.

Zebro project

Advisors/Committee Members: Langendoen, Koen (graduation committee), Verhoeven, Chris (mentor), Nasri Nasrabadi, Mitra (graduation committee), Delft University of Technology (degree granting institution).

▼ 

LoRa is being widely adopted by industrial communities for its long range, robustness and low power wireless communication capabilities. In fact LoRa is gaining more popularity even amongst the common people as it is an affordable solution and operates in the unlicensed radio spectrum. However, LoRa provides a widely heterogeneous coverage; it can reach hundreds of meters or up to tens of kilometers, depending on the surrounding environment. Determining the coverage of LoRa stations is key to provide a good quality of service. On one hand, the traditional method of expensive measurement campaigns can be employed to estimate LoRa's coverage; but this is impractical due to the large geographical areas involved. On the other hand, popular channel models can be adopted; but many of them are yet to be explored for LoRa or rely entirely on the user predening the type of environment to estimate coverage. Neither of those approach are suitable for thousands of non-expert citizens and organizations around the world looking forward to understanding the coverage of their LoRa stations. The aim of the thesis is to automatically estimate the coverage of LoRa, before the deployment of the gateway and without relying on on-site measurements or the user's perception of the environment. Moreover, the estimation must be carried out in a simple, low cost and low eort approach. Considering that the surrounding environment determines in a fundamental manner, the coverage of wireless technologies including LoRa, we use readily available remote sensing information coming from satellites to estimate the characteristics of an area. In this manner, we free up the user from providing any type of data. Based on this remote sensing approach, the thesis provides two main contributions: First we analyze a group of parametric models (ITU-R 1812 and Okumura Hata model) and determine that the Okumura Hata model is better suited for LoRa. Second we improve the performance of using the basic Okumura Hata model by proposing an automated approach that explores remotely sensed height models and land cover maps to automatically congure channel model parameters. The performance is evaluated based on a relative comparison due to some unknown transmitter setting parameters and assess which algorithm accurately tracks the changes in the real path loss. A validation using a relative comparison approach on 18000+ samples of real LoRa data shows that the modied algorithm gives an improved performance compared to the novel approach in path loss prediction and the ITU model. The modied algorithm could improve the coverage up to a factor of 5 compared to the novel approach in free space ranges. Moreover, in an urban built-up city the modied algorithm could improve the coverage by up to 1.5 km compared to the novel approach.

Electrical Engineer | Embedded Systems

Advisors/Committee Members: Zuñiga Zamalloa, Marco (mentor), Langendoen, Koen (graduation committee), Janssen, Gerard (graduation committee), Delft University of Technology (degree granting institution).

▼ 

The idea of having a smart environment to automate day to day tasks appeals to us all. To enable this, we need objects with embedded electronics to communicate with each other in a network known as the Internet of Things (IoT). The IoT communication infrastructure is built on top of existing Radio Frequency (RF) technologies such as Bluetooth Low Energy (BLE), WiFi, and cellular protocols. The RF technology is bandlimited and power hungry, making it unfeasible to support the growing demand of IoT networks. The number of IoT devices in the year 2020 is expected to be around 20 billion, making it essential to explore other areas of sustainable communication technology. Visible light communication (VLC) in the optical domain is being explored to meet the surge in connected devices and to enable sustainability in the energy consumed. The idea behind a VLC system is to toggle a Light Emitting Diode (LED) at high speed to transmit information which ensures that users are not subjected to visual interruptions. Even sunlight - the biggest source of illumination - can be used to transmit information. However, it is not possible to toggle the sun like LEDs. Hence, the objective of this thesis is to use sunlight to setup a green communication channel. In the 1800s, sunlight was used to communicate over long distances by using mirrors to reflect light to send signals. Taking inspiration from this method, I propose using smart materials to toggle sunlight and use it for wireless communication. My aim in this thesis is to analyze the behavior of smart materials, develop a modulation scheme suitable to send information using sunlight, and evaluate the system's performance.

Embedded Systems

Advisors/Committee Members: Langendoen, Koen (graduation committee), Adam, Aurele (graduation committee), Zuñiga Zamalloa, Marco (mentor), Delft University of Technology (degree granting institution).

▼ 

Energy-harvesting devices have enabled Internet of Things applications that were impossible before. Power failures are the norm for these battery-less devices, imposing a challenge to maintain a continued notion of time on these energy restricted devices. To address this challenge we introduce a novel time measurement architecture for energy harvesting devices. Compared to existing solutions, our solution not only improves the start-up time but also reduces the required energy to measure a duration of time.

Embedded Systems

Advisors/Committee Members: Pawelczak, Przemek (mentor), Langendoen, Koen (graduation committee), Litjens, Remco (graduation committee), Delft University of Technology (degree granting institution).

▼ Kite Power Systems present technological advantages over traditional windmills. In this work, the design of a dual wireless link for kite applications requiring high reliability and low delays is presented. The link provides realtime data exchange between the system components that allows researchers to run closed-loop controllers, and to validate kite models. The proposed solution meets the quality of service requirements (delay and packet losses) and other requirements specific for airborne wind energy applications like small form factor, low weight, and power consumption. The most relevant design choices for frequencies, devices, and configuration are described. A pair of controllers for flying crosswind were tested under different QoS conditions with positive results. Performance measurements during flight were used to build statistical models. Such models were implemented in a simulator to better understand the effects of the network over autopilot controllers and to find the minimum communication performance requirement for the system. Advisors/Committee Members: Langendoen, Koen (mentor), Onur, Ertan (mentor), Schmehl, Roland (mentor), Fechner, Uwe (mentor), Delft University of Technology (degree granting institution).

▼ 

The idea of the internet of things (IoT) states that devices with embedded electronics will outnumber the devices operated by humans. All those "things" will require connectivity to exchange information gathered from processed sensor data. A large part of the data will be sent through the air, such that mobile devices can be deployed without a need to connect any wire. Radio frequencies are widely used for wireless communication and include many systems like WiFi, Bluetooth, FM radio stations and mobile telephony. The capacity of radio channels is limited, thus IoT devices might require different methods for communication. In this work we study the behaviour of a relatively new communication channel, called the passive light channel. The passive light channel uses reflections of light to send information in the visible spectrum. It works a bit similar to barcode readers in stores, but it doesn't require a laser. It is sufficient to reflect sunlight or light from other sources, providing a sustainable method of transmitting information. Patterns consisting of black and white parts, or barcodes in particular, are used to regulate the reflected light. A light sensor is used to detect the reflected light and read the information from the reflected light. The key question in this research is how much information we can convey in such manner. In this work we show our three main contributions. First, we show that the passive light channel works with barcodes and reflections of ambient light. Second, a mathematical model and simulation tools are presented that describe the behaviour of the channel. Finally, the performance of the passive light channel with ambient light is analysed using empirical testbeds and compared with results of optical simulations.

Embedded Systems

Advisors/Committee Members: Zuñiga Zamalloa, Marco (mentor), Langendoen, Koen (graduation committee), Al-Ars, Zaid (graduation committee), Delft University of Technology (degree granting institution).

▼ Miniaturization and cost reduction of hardware components have created a trend in the space industry where the traditional centralized computer is being replaced by distributed computer architecture. However, this trend comes with a cost: the on-board software complexity of the space missions has increased. The complexity has origins in the requirements of the missions where in general, these are coordination and control-related processes. As the coordination and the control of the satellite's activities are not trivial tasks, the Multi-Agent Systems(MAS)-approach has been proposed as a new architectural style due to its distributed nature. There are several existing frameworks for implementing MAS-based applications, however, most of them are neither designed to satisfy real-time requirements nor designed to be implemented in highly-constrained embedded systems. Therefore, the purpose of this thesis is to develop a new tool for MAS-based applications: A Multi-Agent Framework for Embedded Systems (MAES). The framework was implemented on top of a Real-Time Operating System: TI-RTOS, therefore, applications implemented with MAES have realtime characteristics. Experiments have shown that the execution time of an Attitude Determination algorithm is consistent on each call with a variance value of the order of 10⁵ [s²], demonstrating the predictability of the framework. Furthermore, the user coding effort is reduced as several routines are standardized and encapsulated into MAES' API. However, the predictability and ease-of-use come with a slight cost: experiments have shown that MAES-based applications lead to an increase of 6.7 KB in average in Flash memory and 4.5 KB in average in SRAM memory with respect to its non-agent implementation. Also, the CPU utilization increases as inter-agent communication requires additional processing time, also increasing the power consumption. However, the increase is low as the results have shown that is less than 1% in average. Advisors/Committee Members: Langendoen, Koen (mentor), Carvajal Godínez, Johan (graduation committee), Menicucci, Alessandra (graduation committee), Spaan, Matthijs (graduation committee), Delft University of Technology (degree granting institution).

▼ 

Nowadays, 19% of the global energy consumption is used for lighting. For this reason, saving energy in lighting is vital. A simple way to save energy is to “simply” turn the lights off, or reduce the amount of light used when nobody is around. This thesis proposes a new method for luminaires to detect the presence of humans and objects which only uses a photodiode and a fraction of the light a luminaire normally emits, namely Dark Sensing. Dark sensing works by sending out short flashes of light. These short flashes use little energy and are barely visible to the user. These flashes get reflected by the environment and received by a photodiode placed next to the light. By extracting a key feature of the received flash, we obtain a metric representing the surrounding area. If an object enters the observed area, the reflections of light will change. These changes will be noticed by the system, which triggers a detection resulting in the light being turned on. A prototype was created which shows the potential of the newly developed method. The prototype was tested in two different environments and achieved detects between 73% and 90% of bypassing pedestrians, depending on the accepted false positive ratio (0 to 0.05).

Embedded Systems

Advisors/Committee Members: Zuñiga Zamalloa, Marco (mentor), Langendoen, Koen (graduation committee), Doerr, Christian (graduation committee), Delft University of Technology (degree granting institution).

▼ Providing more confidentiality by extending a capability-based OS trough a hardware enforced isolation between the memory of the kernel and other memory. By employing memory tagging and hardware based Inter Process Communication (IPC) this defense gives an overhead of 26 percent. Advisors/Committee Members: Wong, Stephan (mentor), van Genderen, Arjan (graduation committee), Langendoen, Koen (graduation committee), Delft University of Technology (degree granting institution).

▼ 

With the arrival of smart electricity meters energy consumption can be monitored continuously and displayed on external devices such as phones or tablets. As a consequence, users become more aware of their energy usage which may result in a reduction in energy consumption. Nevertheless, many countries in Europe still use an analog electricity meter based on a rotating disk, which rotates at a speed proportional to the consumption passing through the meter. In contrast to a smart meter it is not possible to read out the consumption directly. As an alternative, the rotating disk can be observed such that the power (W) and Energy (kWh) consumption can be derived. The current sensor device developed by Quby, uses an LED and phototransistor, where the LED emits light on the disk that is reflected towards the phototransistor. Because of the physical properties of the disk, the sampled phototransistor signal can be represented as a pulse signal, where each pulse indicates a revolution of the disk. The sensor device is mains powered, however, many electricity meters are not located near a power outlet requiring the sensor device to be battery powered. This is a challenging problem since the LED has a relatively large power consumption. The pulse-detection algorithm running on the mains-powered sensor device assumes an LED current of 10 mA with a sampling frequency of 10 kHz. Quby requires that the battery-powered device will last at least one year on an energy budget of 4200 mAh (roughly four AA batteries), and the percentage error of the determined energy consumption should be less than 5%. This implies a factor 50 in power reduction. In this thesis we propose an energy-efficient noise-robust pulse-detection algorithm to detect pulses while keeping the LED current to a maximum of 1 mA. To preserve more energy, the LED is duty cycled to at most 20% instead of 100%, and the sampling frequency is reduced to a maximum of 400 Hz. The proposed method is based on a statistical model where pulse detection is used by means of a multiple-sample likelihood ratio test. Due to the low LED current the signal statistics, such as pulse amplitude, offset and noise, are very sensitive to ambient light. Therefore, an additional method is proposed to estimate these statistics continuously. As a consequence, the detection thresholds in the likelihood ratio test are dynamically adjusted based on predefined probabilities of a false alarm and true detection. The proposed algorithm is extensively tested in a lab setup with three different analog electricity meters, a varying load and a light source emitting light in the same relative spectral-power distribution as sunlight. Moreover, measurements are performed at two households for one week each. From the experiments it can be concluded that the proposed method can last for at least one year when battery-powered, while predicting energy consumptions with an error of less than 2%.

Embedded Systems

Advisors/Committee Members: Taal, C.H. (mentor), van Bakel, W. (mentor), Langendoen, Koen (mentor), Delft University of Technology (degree granting institution).

▼ 

Our society is becoming more and more dependent on communication networks. Thus, network operators and designers need to properly prepare themselves against large-scale disasters that could take down a significant amount of networking hardware within a large area. In this work, we consider the vulnerability of a network to disasters, and we propose an efficient method to compute the distribution of a network performance measure, based on a finite set of disaster areas and occurrence probabilities. Compared to other approaches, our approach is more accurate and gives more information about the vulnerability of the network. Based on our approach, we have implemented a tool to help visualize the vulnerability of a network to disasters. We give general methods to obtain finite disaster sets from empirical disaster data and showcase these methods on actual earthquake and hurricane datasets. Our approach is demonstrated by analyzing the vulnerability of a variety of network topologies to large-scale disasters.

Computer Science

Advisors/Committee Members: Kuipers, Fernando (mentor), Langendoen, Koen (graduation committee), Spaan, Matthijs (graduation committee), Delft University of Technology (degree granting institution).

▼ 

Indoor positioning has several variants as a result of multiple years of study on the topic. Using WiFi signals as the technology to compute the position of the target device is one of the most extended and researched techniques. WiFi Access Points are extensively deployed in all indoor environments where WiFi indoor localization has a potential application. Several of these indoor localization techniques use the Received Signal Strength (RSS) to compute the location of the device. However, the accuracy of the methods that use this parameter is usually low, in the meter range. Eight years ago, the new advances in wireless communications allowed to retrieve the phase of an incoming signal from a commercial chipset, and not only its strength. With the phase, we can use the physical and mathematical principles of signal transmission to compute the distance to a device. In the long term, it can contribute to having a more robust and long-lasting method to perform indoor localization if we can overcome the challenge of obtaining a delay-free phase measurement. For indoor distances, a small delay in the time estimation can be translated into a significant error in the ranging result. In this work, we re-implement the ranging technique of the paper Chronos: Decimeter-Level Localization with a Single WiFi Access Point. This well-known research work explains how to obtain an accurate phase measurement from a WiFi chipset and the technique to compute the distance to the target. We provide a guide of the installation and implementation of the hardware and software required for a system like this, indicating the critical points to allow future researchers a quick set up of the equipment. We revisit the procedure of Chronos to obtain a delay-free phase measurement and show how to re-implement it from scratch. Finally, we perform a validation of the system in four different steps, from the use of ideal data to progressively introducing the real distortions of indoor wireless data. The evaluation of the performance of the system leads us to present the finding of a non-linear delay that can produce significant distance errors.

Electrical Engineer | Embedded Systems

Advisors/Committee Members: Zuñiga Zamalloa, Marco (mentor), Pawelczak, Przemek (graduation committee), Langendoen, Koen (graduation committee), Janssen, Gerard (graduation committee), Delft University of Technology (degree granting institution).

▼ 

Traffic congestion at signalized intersections is a big economical and ecological problem. Handcrafted traffic light controllers (TLCs) are currently used to minimize the impact, but they are expensive to design and maintain and their performance degrades over time. Predictive TLCs and advanced driver assistance systems (ADAS) form a potential solution but are still unfeasible in practice today because of their computational complexity and unpredictability. The distributed predictive TLC developed in this thesis, called DIRECTOR, is feasible and enables time to green/red and green light optimal speed advice (GLOSA) systems. DIRECTOR utilizes predictions of the arriving traffic flows and a model of the current queue length to optimize the traffic light schedule. It can operate in two modes; Ad-hoc mode, where the schedule is generated and applied right away, and fixed-ahead mode, where the schedule is fixed in advance to enable ADAS. DIRECTOR's design makes it scalable and suitable for live learning, eliminating the need for expensive (re)calibrations and improving its performance with more and better data, which will become available in the near future. A long short-term memory recurrent neural network is developed to predict the arriving traffic flows. On a case study this network proves to be on average 4.7% more accurate than the current state-of-the-art model, which is significant for a controller's performance. Simulations of the same case study intersection, which is currently equipped with a state-of-the-art actuated controller with green wave coordination, show that in ad-hoc mode DIRECTOR performs similar to the current controller. DIRECTOR reduces the average delay per vehicle by 1% (from 10.4s to 10.3s) at the cost of an increase of 15% in the average number of stops per vehicle (from 0.40 to 0.46) compared to the current controller. Simulations with ideal predictions show that, in ad-hoc mode, DIRECTOR has the potential to improve the average delay by 8.7% (from 10.4s to 9.5s) while keeping the number of stops equal (at 0.40). Simulations with GLOSA show a 30% reduction in the average number of stops at the cost of a 13% increase of the travel time compared to the ad-hoc mode. Combining this with ideal predictions shows that DIRECTOR in fixed-ahead mode has the potential to keep the average delay equal compared to the current controller, which will greatly improve traffic flow. Compared to a more typical Dutch actuated controller, DIRECTOR achieves a delay reduction of 39% in ad-hoc mode and 23% in fixed-ahead mode. Overall, DIRECTOR is a new data-driven traffic light controller that is relatively easy to set up, reduces costs, can enable advanced driver assistance systems, is futureproof and has the potential to greatly improve traffic flow.

Embedded Systems

Advisors/Committee Members: Langendoen, Koen (mentor), Verhoeven, Eddy (graduation committee), Spaan, Matthijs (graduation committee), Delft University of Technology (degree granting institution).

▼ The e-sport of drone-racing involves human pilots to race against time. Recently, drone races have also gone fully-autonomous. As a result, these agile robotic platforms not only pose challenges of flying fast to the participating pilots but also create challenges for the flight control computers. As a result, the concept of autonomous drone racing has gained significant attention from research groups around the world. These races aim to push the boundaries of perception and control algorithms, while simultaneously mitigating the real-world uncertainty of execution on autonomous systems. While perception algorithms face challenges due to limited feature detection, high motion blur and computational requirements, control algorithms face challenges of convergence to the desired trajectories that are planned out in the race arena. This thesis addresses the challenge of control for racing, which is responsible for guiding the drone to design and track desired trajectories for fast flights. The control sub-modules of racing drones are responsible for generating trajectories for fastest possible flights and also for obeying these generated commands. Additionally, the requirement of limited algorithm complexity is added to match the philosophy of computationally efficient algorithms at the Micro Air Vehicle Laboratory. However, to address the requirements of these control sub-modules, the prerequisite of accurate state estimation always persists. Assigning control actions to a robot without information on the current state of the robot is rather unwise. As a result, this thesis first aims to perform accurate state estimation before designing controllers for time-optimal trajectory tracking. Again, another constraint of using only a single sensor (i.e. the Inertial Measurement Unit) is added to make the drone race in GPS denied environments. As a result, the goal of the thesis is two-fold i.e. making accurate state estimators while using limited sensors and designing optimal controllers for taking the quickest trajectory through the arena. To achieve the goal of accurate state estimation, existing techniques are studied. Several features from each of these methods are selected to design a new estimator. To achieve the goal of time-optimal trajectory generation, firstly, the flaws of traditional control methods are pointed out. A new optimal-control technique is proposed, which makes use of fundamental principles dating back several decades. This principle is then fused along with present-day optimization solvers. Finally, the proposed state estimation and control algorithm are compared against prior (benchmarked) techniques in the area. Compared to existing optimal control techniques, the proposed algorithm leads to faster trajectories and consumes less computational power onboard. Advisors/Committee Members: de Croon, Guido (mentor), de Wagter, Christophe (mentor), Langendoen, Koen (mentor), Zuñiga Zamalloa, Marco (mentor), Delft University of Technology (degree granting institution).

▼ A lot of research is been being done on Visible Light Communication (VLC), which has shown to be of interest for many applications, such as localization. Since localization based on VLC requires active modulation of light sources, this limits the amount of light sources that can be used for localization. Furthermore, in some situations there might not even be a controllable light source present (for example outdoors). To extend the use of light-based localization schemes, this thesis looks into a way to achieve the same result as current VLC localization methods in a passive manner, i.e. without control of the light sources. Previous work has been done on passive ambient light-based localization by Wang et al.: objects are equipped with unique barcodes, that reflect ambient light in a distinct manner. The reflected light is received by photosensors, from which their ID is obtained. However, this work has focused on identifying large-sized objects in one dimension. Using the same principle for localization of small-sized objects, and in two dimensions, are open challenges that this thesis addresses . The work presented here forms a proof-of-concept of a passive light-based localization system for two-dimensional, real-time tracking of small-sized objects. In order to achieve this, a special enclosure has been designed, giving simple photosensors the ability to distinguish small-sized objects without compromising their FOV. With this enclosure, a single photosensor can detect barcodes down to 7 cm in size in the test set-up, while distinguishing up to three different IDs. A particle filter has been implemented to combine detections from different photosensors into a single estimate of an object’s location. The localization system is designed around the robots designed by a MSc student at the Embedded Systems group at TU Delft. By moving these robots at a speed of 15.4 cm/s in a straight line through the test set-up, a localization error of 4.8 cm is obtained. The distance between the robots and the sensor equals 20 cm. Advisors/Committee Members: Zuñiga Zamalloa, Marco (mentor), Pawelczak, Przemek (mentor), de Croon, Guido (graduation committee), Langendoen, Koen (graduation committee), Delft University of Technology (degree granting institution).

▼ 

The performance of gun and rocket propellants, which consist of energetic materials, is largely determined by their geometry and composition. Con- ventional production methods limit the performance by putting constraints on both. With additive manufacturing, or 3D-printing, there are signi- cantly fewer geometry constraints and together with the ability to combine multiple materials into a continuous gradient new performance optimization opportunities are created. In the current 3D-printing world it is possible to print single-material or discrete gradient multi-material objects by trans- lating a CAD model to printer instructions. This translation is done with slicer software that slices a 3D-model and outputs printer instructions in a G-Code le. This thesis looks at how an object with a continuous gradient can be printed. A modied version of the popular Cura slicing software is presented that can apply an approximation of a continuous gradient to an input model. The printer paths are simulated with the slicer software and ultimately printed using TNO's multi-material 3D-printer. While the print results show that energetic materials behave in such a dierent way than normal plastics that 3D-printing them it is not an easy task, printing a 3D-model with a multi-material continuous gradient is certainly viable.

Electrical Engineer | Embedded Systems

Advisors/Committee Members: Langendoen, Koen (mentor), Straathof, Michiel (mentor), Zuñiga Zamalloa, Marco (graduation committee), Rellermeyer, Jan (graduation committee), Delft University of Technology (degree granting institution).

▼ 

The off-side rule is the most complicated rule in soccer. Today, off-sides are determined with the help of referees running on the sides of the field or with the help of cameras. However, for lower league games the installation of a camera based system is very expensive. This work proposes a sensor based system that can be worn by players which can automatically detect off-sides. In order to create such a system, two aspects had to be researched (a) localisation and (b) ball detection. This work provides an evaluation of an ultra-wide band system in outdoor conditions to solve the localisation problem and various proximity based sensors to solve the ball detection problem. The primary research objective of this work was to find the pit-falls of the ultra-wide band system with respect to our application. It was found that the tri-lateration algorithm implemented on the devices contributed to a lot of errors in localisation accuracy and the overall system suffered from significant packet loss. A custom tri-lateration algorithm along with hardware improvements are proposed. Our results show that the localisation errors are reduced by 56 % and improved system suffers from nearly no packet loss. As for ball detection, results show that ToF sensors provided the best accuracy in detecting the ball.

Embedded Systems

Advisors/Committee Members: Langendoen, Koen (mentor), Zuñiga Zamalloa, Marco (mentor), van Genderen, Arjan (mentor), Delft University of Technology (degree granting institution).

▼ 

Artificial lights in our buildings have always been designed to be static for one purpose only; illumination. With the increasing interest in Visible Light Communication (VLC) and a crowded radio-frequency spectrum, static lights in our offices and rooms might not be ideal anymore. VLC can add extra services like high quality communication and localization. These services benefit from optimal circumstances as we can find in the hot spot below the lights. Outside the hot spot, the intensity of the lightbeam starts to drop rapidly. This research is dedicated to the aspect that the static lights are lacking which is mobility. Instead of having static lights we create a light source that is able to move along with persons in a room. In this work, which consists of the design and evaluation of our system, we create a mobile platform which includes enhancement of an existing gondola system, the design of a small mobile LED-transmitter and a smartphone as receiving device. We designed a system for the light to follow a user. The smartphone can locate the light and it calculates the location offset through image processing. New coordinates for the light are send to the gondola system to move the light and keep the transmitter and receiver in sync. The system was tested by installing it in the Embedded Systems laboratory of the TU Delft and evaluated in terms of illumination, communication and energy efficiency compared to a static light network. We can conclude that a mobile light can add extra services to VLC by ensuring a constant high illuminance and signal-to-noise ratio by using individual focused hot spots that follow the user in real-time, with an average position error of 1.74 cm, which the static light network can not do causing a performance degradation. The mobile light is in most cases also more energy efficient than a full hot spot static light network coverage. Besides that, we obtain a 3.91 centimeter precision indoor localization as an extra result.

Electrical Engineer | Embedded Systems

Advisors/Committee Members: Zuñiga Zamalloa, Marco (mentor), Langendoen, Koen (graduation committee), Lofi, Christoph (graduation committee), Delft University of Technology (degree granting institution).

▼ Daylighting is the immediate exploitation of solar energy in the form of nat- ural lighting and plays an integral role in minimizing the energy footprint of a building. Smart daylighting enables us to design buildings that provide comfort and energy savings. This work proposes a dynamic facade system for buildings which aims to maximize user comfort while simultaneously maximizing energy savings by harvesting solar energy optimally. The solar panels on the facade can har- vest the highest amount of energy when it is positioned perpendicular to the suns rays. However, this may result in unsatisfactory lighting conditions in- side the room and the problem is approached as a bi-objective optimization problem. This work is a preliminary exploration of the concept of smart skins for buildings that autonomously regulates light while harvesting solar energy, contributing to the creation of the future of sustainable buildings. The primary focus of this research work revolves around building a con- ceptual model, formulating an optimization problem, developing a control algorithm, iFOS, and then evaluating it. Data was simulated using advanced simulations to evaluate the dynamics of light indoors. Two benchmarks were created to evaluate the algorithm against, one where the system works towards maximizing user comfort indoors, and the other, where the system works to maximize the energy harvested by the facade. Upto 8% increase in the energy harvested was achieved with minimal loss in user comfort in the use case evaluated. The average energy figure for The Netherlands in the summer months is about 5 kWh/m2/day, which makes the total energy that can be captured at 20% efficiency to be about 750 kWh per day. The algorithm is found to work the best when the desired light level to be maintained indoors lies in the range [400,600] lux. Advisors/Committee Members: Venkatesha Prasad, Ranga Rao (mentor), Sarkar, Chayan (mentor), Langendoen, Koen (graduation committee), Eisemann, Elmar (graduation committee), Delft University of Technology (degree granting institution).

▼ 

Most wireless communication technologies have been using the Radio Frequency (RF) spectrum for decades. Due to the popularity of the Internet of Things (IoT), the RF spectrum has started to become densely populated. Researchers have begun to explore other bands of the electromagnetic spectrum that can be utilized as a communication media. One of the promising choices is the visible light spectrum. Visible Light Communication (VLC) refers to the wireless communication technology that utilizes the visible light spectrum. This spectrum is thousands of times wider than the Radio Frequency (RF) spectrum and is license-free. In VLC, data is transmitted by turning a light source on and off. However, not every light source can be controlled. Passive light sources such as the sun provide an immense amount of light that can be used for wireless communication if we can develop ways to modulate them. One of the researches that use ambient light to create a wireless link is LuxLink. LuxLink uses liquid crystal shutters to control passive light sources and provides low energy, reliable, and flicker-free (safe) communication. This thesis addresses several problems that the current LuxLink system has. We present LuxLink+, an extension of LuxLink that provides two main improvements. Firstly, the data rate of the system is relatively low (80~bps). To increase the data rate, we provide a thorough analysis of the system’s bandwidth. Afterward, we modify the modulation technique, which increases the data rate to 1000~bps at a range of 1.5~m. Secondly, the system has a static data rate, which means that the system cannot adapt its data rate to changes in the environment. We implement a rate adaptation algorithm that can change its data rate accordingly. LuxLink+ improves the average throughput of the system by up to 85 percent compared to LuxLink.

Embedded Systems

Advisors/Committee Members: Zuñiga Zamalloa, Marco (mentor), Langendoen, Koen (graduation committee), Al-Ars, Zaid (graduation committee), Delft University of Technology (degree granting institution).

▼ Progress has been made recently in the research of equal hierarchy backscatter communications. There are, however, challenges to overcome before this technology can become mainstream. For example, state-of-the-art solutions to the phase cancellation problem that occurs in such networks are still trivial and sub-optimal, especially in terms of resource utilization. This thesis aims to address the issue by designing a novel networking algorithm. This is done by (i) characterizing and analyzing such networks to assess the potential of the proposal, and (ii) experimentally demonstrating the concept in the largest implemented network so far. Findings show that the solution presented in this work not only matches current phase cancellation solutions in terms of robustness, but also extends communication range and enables connecting networks served by separate exciters. Advisors/Committee Members: Pawelczak, Przemek (mentor), Langendoen, Koen (graduation committee), Bozzon, Alessandro (graduation committee), Majid, Amjad (graduation committee), Delft University of Technology (degree granting institution).

▼ 

An Indoor Positioning System (IPS) is being developed at TOPIC Embed- ded Systems to track equipment in hospitals. The system should prevent the loss of equipment en make procedures more efficient. The IPS will con- sist of anchors and tags. Anchors are the radios that form an infrastructure in the building to localise tags that are placed on equipment. Different localisation techniques and methods exist for indoor localisation, of which Ultra-wideband (UWB) is a very promising technology as it is robust to mul- tipath interference. Also the Time Difference of Arrival (TDoA) method has advantages over Two-way Ranging (TWR) in terms of energy consumption of a tag and the rate of supported location measurements. A fundamental requirement for Time Difference of Arrival (TDoA)-based localisation is that all anchors must be precisely synchronised as radio signals propagate trough air with the speed of light. Existing synchronisation solutions synchronise the anchors either trough wires or wirelessly. To keep the installation costs of the IPS to a minimum synchronisation should work without adding extra infrastructure to a building such as a clock distribution network. Therefore a wireless solution is required. The existing solutions have been evaluated on hardware equipped with precise clock sources that have tight tolerances. These clock sources are not available on the com- modity hardware (Decawave DWM1001) that is intended to be used by TOPIC. Also the existing synchronisation algorithms are not designed for large multi-hop networks. A new synchronisation algorithm based on a 3-state Kalman filter is devel- oped and evaluated with the existing solutions showing that linear interpola- tion performs the best in terms of the Mean Absolute Error (MAE),Ḣowever the linear-interpolation algorithm comes at the cost of a latency as the times- tamps become only available after the next synchronisation message. If the latency (order of seconds) cannot be tolerated, the developed 3-state Kalman filter is the best alternative. As TOPIC requires a latency of 5 min the linear-interpolation algorithm is integrated in a synchronisation scheme for multi-hop networks. This scheme has the additional ability to measure the propagation delay. The linear- interpolation algorithm is evaluated with practical experiments for single- hop and multi-hop synchronisation. The MAE of the synchronised clock is 229 ps for single hop and 258 ps for multi hop when using a synchronisation period of 1 s. This shows that that the synchronisation algorithm is very suitable for multi-hop networks. To determine the effect of the clock synchronisation on position accuracy an experiment has been conducted where anchors measure the TDoAs of a message sent by a tag. A standard multilateration algorithm [1] was usedto estimate the locations based on these measurements. The accuracy of an individual estimate was low, but by averaging subsequent measurements a mean error of 51 cm was achieved, meeting the requirement of 1 m accuracy.

Electrical Engineer |…

Advisors/Committee Members: Langendoen, Koen (mentor), Leus, Geert (graduation committee), Van Den Heuvel, Dirk (mentor), Delft University of Technology (degree granting institution).

▼ 

Batteryless network nodes could enable unprecedented ubiquitous sensing applications, while gathering energy from the environment and hence removing maintenance costs for batteries. However, these devices inherently suffer frequent power failures and reboots, due to the volatility of the incoming energy, thus classical synchronous message passing cannot work. On the other hand, the asynchronous scheme is not viable either, since power outages can occur at different rates on two devices, and for unpredictable lengths of time. In order to align wake-ups of wireless nodes, a non-canonical mechanism to keep track of time must be devised. We present a novel architecture for reliable packet exchange between energy-harvesting devices, resilient to power failures and agnostic about the physical layer. The proposed architecture, implemented on real hardware, shows an improvement in throughput of a factor of 10, and reduces energy waste by 33 % to 78 %, as compared to the asynchronous message passing.

Embedded Systems

Advisors/Committee Members: Pawelczak, Przemek (mentor), Langendoen, Koen (graduation committee), Erdweg, Sebastian (graduation committee), Delft University of Technology (degree granting institution).

▼ Energy harvesting and battery-free sensing devices show great promise for revolutionizing computing in every known area while expanding to non-conventional use-cases. The promise of cheap, dense, and ubiquitous sensing technology brings new applications for the Internet of Things. However, the future programming model is blurry and complex. With a potential for trillions of devices, and thousands of devices per person on earth, programming languages and associated operating systems must be usable, flexible, and resource efficient. Although transiently powered computing is an area of ample research, no model presented so far has been widely adopted, hindering widespread use. Because of the thousands of applications and differences in requirements, a kernel that abstracts the intricacies of intermittency may be a part of the solution. This thesis explores key concepts that push intermittent systems closer to traditional embedded programming while examining resources costs, feasibility, and motivation for a kernel for intermittent systems. Advisors/Committee Members: Pawelczak, Przemek (mentor), Langendoen, Koen (graduation committee), Finavaro Aniche, Maurício (graduation committee), Delft University of Technology (degree granting institution).