Electrical and Electronic Engineering - Theses
Permanent URI for this collection
Search Results
1 - 3 of 3
-
ItemFinite-time algorithms and performance bounds for real-time Internet of Things( 2019)Rapid developments in technology have enabled a large scale deployment of interconnected sensors and actuators, captured under the umbrella term Internet of Things (IoT). Real-time IoT applications in smart grid, smart traffic control etc. are made possible by the real-time processing of high volume data, generated from dedicated and multi-purpose sensors, and exchanged over heterogeneous wired or wireless communication networks. Our work focuses explicitly on smart intersection management applications and develops and analyses algorithms to cater to its stringent latency, mobility and geo-distribution requirements. Considering the communication delays, distributed IoT implementations like these prefer fog/hybrid architecture-based data processing to the conventional centralised and cloud-based approach. Further, for relevant distributed real-time IoT algorithms, finite-time performance matters more than the asymptotic results in the literature . Thus, precise estimates have to be obtained on the delay needed for an optimisation algorithm to compute a solution within the desired proximity of the optimal solution. Such trade-offs are inevitable for the design of real-time algorithms over an IoT network. This thesis develops distributed optimisation algorithms, which involve explicit delay-accuracy trade-offs possible and studies the effects of channel impairments and communication network structure on them. We introduce a finite-time distributed optimisation algorithm and derives universal performance bounds for an asymptotic algorithm solving a quadratic Network Utility Maximisation (NUM) problem using quantised inter-agent communication. The finite-time algorithm is then used to solve a Model Predictive Control (MPC) problem and applied to a smart traffic intersection management scenario.
-
ItemOptimization and deep learning techniques for next-generation wireless communication networks( 2019)Due to the explosive growth of consumer electronic devices, such as smartphones, tablets, and the Internet of Things, the global mobile data traffic is estimated to increase seven fold by 2022 in the fifth generation (5G) of wireless communication networks. At the same time, the mobile network connection speed is envisioned to increase more than three-fold by 2022. Many technologies have been proposed to fulfill such unprecedented user demands. On the macroscopic level, the network architecture largely determines the performance of the network. Novel network architectures, such as heterogeneous networks (HetNets) and centralized radio access networks (C-RANs), have been proposed to accommodate massive numbers of wireless devices. On the microscopic level, the control of all devices is vital for network operations on a daily basis. Intelligent control agents are hence required to operate networks without human intervention. In this thesis, we start off by focusing on the architecture side of network and investigate a MINLP problem of a joint backhual-access HetNet by using a classical optimization approach. Then, we move on to the operational side of networks and focus on spectrum sharing in cognitive radio networks and topology control in wireless sensor networks. In these problems, we employ deep learning approaches that can learn from collected data and adapt to the changing radio environment without a priori knowledge about the network. We show the applicability and superiority of our deep learning-based algorithms compared with classical analytic approaches. More importantly, we show the novel applicability of deep learning in solving MINLP problems that are commonly encountered engineering problems in wireless communication networks.
-
ItemPrivacy-preserving machine learning and data aggregation for Internet of Things( 2018)The proliferation of Internet of Things (IoT) devices has contributed to the emergence of participatory sensing (PS) and collaborative learning (CL), where multiple participants collect and report their data to a cloud service to analyse the union of the collected data in the server-based framework. While in the decentralized framework, multiple participants collaboratively train a more accurate global model or multiple local models. However, the possibility of the cloud service or any participant being semi-honest or malicious pose a serious challenge of preserving the participants' privacy. Privacy-preserving machine learning and data aggregation aim to discover or derive useful statistics without compromising privacy. This thesis systematically investigates state-of-the-art techniques for privacy-preserving machine learning and data aggregation in a range of IoT applications. Extensive theoretical and experimental results are provided to support the following primary contributions. First, we explore three privacy-preserving machine learning applications. Examples include collaborative anomaly detection, human activity recognition and decentralized collaboration in a biomedical domain. We tackle security challenges in collaborative anomaly detection with a two-stage scheme called RG+RT: in the first stage, participants individually perturb their data by passing through a nonlinear function called repeated Gompertz (RG); in the second stage, the perturbed data are projected to a lower dimension using a participant-specific uniform random transformation (RT) matrix. The nonlinear RG function is designed to mitigate maximum a posteriori (MAP) estimation attacks, while random transformation resists independent component analysis (ICA) attacks. For human activity recognition, a similar two-stage scheme called RG+RP is proposed, the difference lies in the second stage, where participants project their perturbed data to a lower dimension in an (almost) distance-preserving manner, using a random projection (RP) matrix. The random projection can both resist ICA attacks and maintain model accuracy. These proposed two-stage randomisation schemes are assessed in terms of their recovery resistance to MAP estimation attacks. Preliminary theoretical analysis as well as experimental results on synthetic and real-world datasets indicate that both RG+RT and RG+RP exhibit better recovery resistance to MAP estimation attacks than most state-of-the-art techniques, meanwhile high utility is guaranteed. To mitigate the inherent limitations in the centralized framework, and investigate the applicability of the decentralized framework, we study the decentralized collaboration in a biomedical domain. In particular, we develop an efficient Decentralized Privacy-Preserving Centroid Classifier (DPPCC) considering three practical scenarios, where distributed differential privacy (DDP) is combined with distributed exponential ElGamal cryptosystem to preserve privacy and maintain utility. We realize DDP using discrete Gaussian mechanism without any restriction on ε as in the traditional Gaussian mechanism, and only the encrypted noisy model parameters or test results are shared among all parties. It ensures each party learns nothing but the noisy sum of local statistics. Second, we examine privacy-preserving data aggregation in smart grid application. To this end, we propose a multi-level aggregation framework based on fog architecture, which