This thesis addresses the design and optimization of sparse non-uniform optical phased arrays (OPAs) for advanced automotive LiDAR systems. As autonomous driving technologies advance, the demand for high-resolution, reliable, and compact LiDAR systems has become increasingly critical. Traditional uniform OPAs, while effective, face limitations regarding power consumption. This work introduces an innovative approach to designing sparse non-uniform OPAs that achieve desired performance metrics essential for automotive applications, including beamwidth, field of view, and sidelobe levels, while minimizing element count and, consequently, energy consumption.

Through mathematical modelling and simulation, we formulate the problem of sparse OPA design as an optimization problem, leveraging techniques from compressive sensing to identify the most efficient element arrangements. We propose using the sparse array synthesis method to formulate the sparse OPA design problem, utilizing algorithms such as LASSO, thresholding, and iterative reweighted l1-norm minimization to achieve optimal sparse configurations. Our results demonstrate substantial improvements in effectiveness, offering a practical solution to the constraints posed by current LiDAR systems. This thesis contributes to the field by providing a comprehensive framework for the design of sparse non-uniform OPAs, highlighting the trade-offs and benefits of various design strategies. The findings advance our understanding of OPA design principles.

Current methods in Federated and Decentralized learning presume that all clients share the same model architecture, assuming model homogeneity. However, in practice, this assumption may not always hold due to hardware differences. While prior research has addressed model heterogeneity in Federated Learning, it remains unexplored in fully decentralized or peer-to-peer settings. Therefore, in this paper, we investigate a real-world yet challenging situation involving model heterogeneity in a fully decentralized context. Furthermore, we introduced a Model Agnostic Peer-to-peer Learning (MAPL) framework, which allows simultaneous learning of heterogeneous personalized models. Additionally, we define a graph learning objective to infer optimal collaboration weights based on task similarity. Experiments reveal that even in this challenging scenario, MAPL delivers competitive results while being communication efficient owing to the sparse collaboration graph in both model homogeneous and heterogeneous settings.

Future communication scenarios will require massive Multiple-input multiple-output (MIMO) by the use of multi-beam antenna systems. Maximizing the number of beams for a given antenna size is paramount given that the space allocated to the antenna is often limited. This work aims at evaluating the maximum number of beams by analyzing the SIR in different communication scenarios for planar antenna structures. The concept of ‘observable field’ is used to quantify the power received from the desired signal as well as the power associated with the interference. Due to the planarity of the considered antenna structures, the radiating domains introduce scan loss, an effect not previously modelled when considering earlier investigations based on spherical domains. Furthermore, methods of reducing interference in order to improve the SIR were investigated, i.e., the use of a tapered current distributions on the radiating apertures and null placement techniques.

The versatility to learn from a handful of samples is the hallmark of human intelligence. Few-shot learning is an endeavour to transcend this capability down to machines. Inspired by the promise and power of probabilistic deep learning, we propose a novel variational inference network for few-shot classification (coined as TRIDENT) to decouple the representation of an image into semantic and label latent variables, and simultaneously infer them in an intertwined fashion. To induce task-awareness, as part of the inference mechanics of TRIDENT, we exploit information across both query and support images of a few-shot task using a novel built-in attention-based transductive feature extraction module (we call AttFEX). Our extensive experimental results corroborate the efficacy of TRIDENT and demonstrate that, using the simplest of backbones, it sets a new state-of-the-art in the most commonly adopted datasets miniImageNet and tieredImageNet (offering up to 4% and 5% improvements, respectively), as well as for the recent challenging cross-domain miniImagenet --> CUB scenario offering a significant margin (up to 20% improvement) beyond the best existing cross-domain baselines.

Distributed formation control has received increasing attention in multiagent systems. Maintaining certain geometry in space is advantageous in many applications such as space interferometry and underwater sensing. At present, there is a variety of distributed solutions for agents to converge to desired formations and track a series of prescribed maneuvers. They typically rely on the relative kinematics e.g., relative positions of the neighboring agents as state observations for the local controller. In harsh working environments, the acquisition of the relative kinematics is challenged and observation losses might occur, which can be detrimental to the optimality of formation.

In this work, observation losses in noisy environments are addressed under a distributed formation control framework. Three types of solutions are proposed to enhance the robustness which is evaluated through the improvements of tracking error, convergence speed, and smoothness of trajectories in both random and permanent loss settings. Firstly, a relative localization technique is proposed using formation itself as a spatial constraint. Secondly, a dynamic model is established for the agents entailed by a Kalman filter-based solution. Finally, a fusion of the previous two types is inspired and it exhibits superior performance than both aforementioned types individually.

This work not only provides means of relative localization without additional sensor data but also shares insights into coping with random or permanent graph changes for stress-based formation control systems. This could potentially lead to the exploration of formation control with subgraphs or energy-efficient sensing as future directions.

In recent years, the large increase in connected devices and the data that is collected by these devices has caused a heightened interest in distributed processing. Many practical distributed networks are of heterogeneous nature. Because of this, algorithms operating within these networks need to be simple, robust against network dynamics and resource efficient. Additionally, if privacy preservation methods are properly implemented, they add to the power of distributed processing by making it possible to leverage the data of many different users, without infringing the privacy of the individuals involved.
In this study we focus on the primal-dual method of multipliers (PDMM), which is a promising dis- tributed optimisation algorithm that seems to be suitable for distributed optimisation in heterogeneous networks. Most theoretical work that can be found in existing literature focuses on synchronous ver- sions of PDMM. However, in heterogeneous networks, asynchronous algorithms are favourable over synchronous algorithms. So far, simulation results have indicated that asynchronous PDMM converges and can even converge in the presence of transmission losses.
In this work we analyse the properties of stochastic PDMM, which is a general framework that can model variations of PDMM such as asynchronous PDMM and PDMM with transmission losses. We build upon previous empirical results of PDMM and formulate theoretical proofs to substantiate these results. After defining stochastic PDMM and proving its convergence, we compare a number of PDMM variations that have been mentioned throughout the literature. Lastly, we derive a lower bound for the variance of the auxiliary variable in the context of stochastic PDMM, assuming uniform updating probabilities. This lower bound indicates that subspace based privacy preservation is applicable to certain instances of stochastic PDMM, like asynchronous PDMM.
The main result of this work is a theoretical proof that shows that stochastic PDMM converges almost surely if the updating probabilities of each auxiliary variable are nonzero. Two important conclusions that follow from this proof are the almost sure convergence of asynchronous PDMM and unicast PDMM with transmission losses. Another useful result is the fact that subspace based privacy preservation is effective when using asynchronous PDMM.

Formation control problems consider a set of mobile agents with the underlying goal of attaining and maintaining a state where the relative positions of agents are stable in accordance with the desired configuration.
Navigation for formation control is typically achieved through localization in a global reference frame, e.g., via GNSS. However, when a global reference frame is not shared among agents, a relative navigation approach is required.
Distributed filtering for relative localization in formation control systems is a relatively unexplored field.
The absence of absolute positioning means motivates the need for a distributed filter that operates on the edges of the sensing graph of the multi-agent system.
In this thesis, a data model for relative formation control problems and two edge-based Kalman filters are proposed. The first filter is designed for an individual edge. The second is a filter designed via decoupling of the optimal global filter which allows for the joint estimation of adjacent edges. It is shown that the joint filter is optimal under the decoupling constraints.
Monte Carlo results show that when random environmental disturbances are correlated among agents, the joint filter outperforms the local edge filter in a mean square error sense.
Lastly, systems are considered where inter-agent communications are unavailable, leading to biased prediction steps of the Kalman filters. We aim to minimize this effect through the proposal of a local Wiener filter which predicts the control actions of neighboring agents.

Time-varying network data are essential in several real-world applications, such as temperature forecasting and earthquake classification. Spatial and temporal dependencies characterize these data and, therefore, conventional machine learning tools often fail to learn these joint correlations from data. On the one hand, hybrid models to learn from time-varying network data combine several specialized models able to capture these dependencies separately, thus ignoring their joint spatio-temporal interactions. On the other hand, state-of-the-art approaches for jointly learning time-varying network data do not exploit the useful prior provided by a graph-time product graph. This prior structural knowledge can aid learning and help the models improve their performance. For this reason, we propose a novel neural network architecture to learn from time-varying network data using product graphs and graph convolutions, thus exploiting this prior during learning. In particular, our architecture (i) learns the most suitable graph-time product graph to represent the time-varying network data and model the graph-time interactions; (ii) performs graph convolutions over this product graph to learn the graph-time interactions from data; (iii) employs graph-time pooling to reduce the dimensionality over layers. To the best of our knowledge, no research has yet attempted to capture spatial and temporal dependencies using graph convolutions over product graphs. Results on synthetic and real-world data show that the proposed method is useful in learning from time-varying network data for both regression and classification tasks. For classification, we cure a real-world dataset for earthquake classification and compare the proposed approach with state-of-the-art models. Results indicate that graph-aware models outperform graph-unaware models on this task. For regression, we evaluate the proposed method on two real-world temperature datasets and compare our architecture with graph-aware and graph-unaware models. Results on the smaller dataset show that linear models outperform neural-network models. On the larger dataset, we find that prior knowledge about graph-time interactions seems to be less beneficial in case of abundance of data. For the synthetic experiments, we find that (i) learning the structure of the graph-time product graph from data improves the performance compared to adopting a fixed type of product graph; (ii) learning sparser graph-time product graphs further improves the performance; (iii) the proposed graph-time pooling technique contributes to the model's generalization capabilities.

People counting data in offices is used in many applications like HVAC system control and space management to increase comfort, decrease energy consumption and optimise space utilisation. In contrast to past approaches using imaging modalities that tend to be either expensive or intrusive, we consider single-pixel thermopile sensors for people counting. These sensors may already be deployed as part of a smart lighting system to provide temperature data for HVAC controls. Firstly, a statistical sensor model for thermopile temperature measurements is proposed. The proposed people counting method enhances the CUSUM RLS algorithm to estimate temperature change caused by people entering or leaving. We estimate mean temperature changes upon detection of an occupancy event, and then estimate the people count using a maximum likelihood on the estimated temperature change. Finally, PIR vacancy data is merged with the people count estimation to increase accuracy. We obtain an average counting error of 0.11 and 0.19 for 90% of the instants respectively when considering 15 minute windows for simulated and experimental datasets. A second aspect of the thesis considers the problem of commissioning plan detection. We leverage the two-sided CUSUM signals to address this problem. The two-sided CUSUM scores for a pair of sensors are used to calculate similarity measures; these features are used in a Random Forest Classifier to detect commissioning changes of the sensor pair. Using simulated data with the thermopile signal model, we show that the proposed method achieves a true positive rate (determining the correct layout) of 90.2% and false positive rate of 1.3%.

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.

Automotive radars play a very important role in reduction of traffic accidents and casualties by making the vehicle fully self aware of its surroundings. For the vehicle to be fully self aware multiple sensors or radars have to placed in close proximity of each other on the body of the automobile. This will lead to interference between the radars which results in sub optimal performance of the radar system. Currently most of interference mitigation techniques employ various methods of signal coding algorithms, by assigning each radar a different code or modifying beamforming and angle of arrival algorithms to suppress the interference. Most of theses technique use and exploit the properties of the transmitted signal (signal processing techniques) for mitigating the interference. In this thesis the avenue of antenna array design for mitigation of interference is explored. By modifying the radiation pattern of the array, the antenna array will act as spatial filter against the incoming interference signals. In this thesis two techniques for designing the antenna array with improved radiation pattern are proposed. The design techniques are used to optimize the MIMO virtual array of the radar for low sidelobes and increased aperture, and the optimized virtual array is then factorized to obtain the Tx and Rx arrays. 2 virtual arrays of 12 elements and 24 elements are optimized using the design techniques. The performance of the various designs is ranked based on the output SINR of an adaptive beamformer. The simulations results show that the optimized MIMO arrays result in an improvement of up to 3.5dB in SINR in comparison to the currently used topologies. Finally, a prototype of the improved MIMO antenna system is modeled in CST MS design environment, and the superior performance of the designed array topologies over the existing array topologies is verified.

Automotive radars play a crucial role in the reduction of traffic casualties and the realization of autonomous driving due to its robustness and adverse weather tolerance. However, as the penetration rate of automotive radars increases, concerns arise regarding the mutual interference. In contrast to the predominant Frequency-Modulated Continuous Wave (FMCW) radars, Phase-Modulated Continuous Wave (PMCW) radars might provide solutions for the interference problem by coding the waveforms, which brings an additional degree of freedom. In this thesis, a system level overview of the phase-coded radar is presented. Regarding the popular code families, particular attention was given to the Gold, Almost Perfect Autocorrelation Sequences (APAS), and Zero-Correlation Zone (ZCZ) sequences. This thesis has proposed three distinct designs for 16-TX-element MIMO Phase-Coded radar that drives the requirements of a Medium Range Radar (MRR). The conclusion can be drawn that APAS and ZCZ sequences have shown to provide a (sub)optimal design that is emphasized by their perfect correlation characteristics within the designated distance of interest. Furthermore, in full-transmit capacity MIMO, the proposed designs use semi-orthogonal waveforms to provide excellent performance in synchronous PMCW radars. However, performance degrades in case Doppler-shifted reflections are received, whereas sidelobes arise in range due to non-orthogonality. This thesis concludes that the APAS-coded waveforms are most tolerant to Doppler shift; besides, its Time Division Multiplexing (TDM) MIMO Transmission scheme is computationally more efficient than the Code Division Multiplexing (CDM) scheme used in Gold and ZCZ coded waveforms. Finally, the radar-to-radar interference investigation has proven that for uncorrelated PMCW- to-PMCW interference, the noise floor undergoes a consistent increase, which is according to the interference-plus-noise power level calculated from the link budget analysis. This thesis emphasizes these results, as the interference effect in FMCW-to-FMCW (provided by NXP Semiconductors) have seen similar phenomena. Therefore, results in this thesis support the conclusion that coded waveforms do not remove or suppress the energy levels of the interference. For correlated interference, ghost targets might be formed, but this depends on the starting time as well as the starting coded bit of the interferer, as the correlation peak can be formed outside the distance of interest.