Search

Transformative sensing is a new research trend that seeks to cross-fertilize sensing functionality in other ubiquitous pieces of electronics, thereby achieving pervasive sensing with ultra-low overhead. Examples include detecting human activity with kinetic energy harvesters embedded in wearable devices, recognizing hand gestures with Wi-Fi, and so on. This thesis makes three new contributions to transformative sensing, which involve gait detection, gesture recognition, and communication channel estimation. In the first contribution, the thesis devises solutions that enable use of the same wearable PEH for both energy harvesting and gait recognition at the same time. This is achieved by designing a special algorithm that minimizes distortions caused to the sensing signal when the harvested energy is being stored in the capacitor. The concept is prototyped in the form factor of a shoe, and its performance is evaluated with 20 subjects, which confirms that the proposed filter helps detect human gait with 8% higher accuracy while consuming 35% less power compared to the state-of-the-art. The second contribution is the investigation of gesture recognition using solar cells, where the power signals from solar cells exhibit distinct patterns under different gestures. To improve the system robustness under different settings, a signal processing pipeline was devised by combining z-score transformation and DTW. Experiments with both transparent and opaque solar cells confirm 96% gesture recognition accuracy. The third contribution investigates the use of inertial sensors for CSI acquisition in the context of vibration-based MIMO communication over human skin. It is first demonstrated that due to slow ramping of vibration motors, conventional channel sounding based CSI acquisition is not practical for vibration MIMO. To solve this problem, this thesis proposed a deep learning-based CSI acquisition framework to accurately predict CSI entirely based on inertial sensor (acceler

Innovative sensing modalities for wearable devices continue to emerge with the advancement of microelectromechanical systems and material science. Compared to conventional sensing modalities, these are either more accurate or more energy efficient. In keeping with this trend, this thesis introduces innovative sensing modalities, aiming to quantitatively improve on accuracy or energy consumption and finally lead to qualitatively difference in working scenarios. The contributions of this thesis are three-folded. Firstly, via implementing a new type of highly sensitive, stretchable, optical transparent, and low-cost strain sensor, we addressed the problem of human posture detection with casual loose-fitting smart garments. We developed a smart garment system—E-Jacket—by attaching novel sensors to an off-the-shelf casual jacket. In addition to the novel highly sensitive strain sensor, we implemented a state-of-the-art human activity recognition tool, the combination of a convolutional neural network and a long short-term memory recurrent network. Our evaluation showed that the E-Jacket can achieve up to 91.7, 84.4, and 88.4 per cent detection accuracy, respectively, for these three case studies. Second, we developed H2B, which is a system for securely pairing wearable devices by generating a shared secret key from interpulse interval between heartbeats. In the H2B system, we implemented cost- and energy-efficient piezoelectric sensors to replacing conventional electrocardiogram or photoplethysmogram sensors. To handle the high rate of bit mismatch, we also proposed a novel compressive sensing-based reconciliation method that can be adopted on all kinds of symmetric key generation systems. Our results showed that H2B has a pairing success rate of 95.6 per cent. Third, we explored the feasibility of using KEHs to collect gait signals for device authentication and symmetric secret keys generation continuously. Based on preliminary analysis, we developed KEHKey, a gait-base

Millimeter wave networks promise to offer ultra-fast internet download speed, but the access points or base stations must always align the beams precisely to client devices. Efficient beam alignment for mobile users therefore is considered one of the most challenging problems facing millimeter wave networks. Existing approaches that use in-band beam alignment suffers from long alignment delays and low communication performance, especially when large number of mobile clients are connect to the access point. In this research, we explore the benefits of out-of-band inaudible sound assisted beam alignment to reduce the outage probability, thereby improving the performance gain of antenna in millimeter wave beamforming. In particular, this thesis makes three fundamental contributions. First, we analytically study the beam alignment performance of 802.11ad in the presence of multiple devices while rotating with an applicable angular velocity. We come up with a probabilistic model for required number of beacon intervals to complete antenna training in multi-users scenario for 802.11ad. Second, we propose to take advantage of inaudible sound as a side channel to detect the direction of client and assist beam alignment in millimeter wave access points. Using a combination of experimental and simulation analysis of the inaudible sound spectrum available in typical mobile phones, we demonstrate that the use of 50 Hz and 50 ms sound chirps with an array of two microphones provide efficient and reliable detection of direction. Moreover, we design a filtering approach using FDM channel access to correctly assign the sound source corresponding to the estimated angle on the receiver side. Third, we conduct a comprehensive simulation in order to evaluate the performance of the proposed sound assisted beamforming on the gain of antenna. \deleted{Initially, the proposed analytical model is validated by the developed simulation platform.} We show that our proposed algorithm achieves a

Advancements in nanotechnology promise new capabilities for the Internet of Things (IoT) to monitor extremely fine-grained events with sensors as small as up to a hundred nanometers. Researchers predict that such tiny sensors can be connected to the Internet using graphene-based nano-antenna radiating in the terahertz band, giving rise to the so called Internet of Nano-Things (IoNT). Powering such wireless communications with nanoscale energy supply, however, is a major challenge to overcome. Since in many application domains, different types of events discharge different amounts of energy to the environment, we propose an energy-neutral event monitoring framework, called eNEUTRAL IoNT, that allows the sensors to transmit event information using only the amount of energy harvested from the events. We design and analyze two implementation methods for this framework. The first method uses a single pulse containing the entire energy harvested from the event but manipulates its pulse width (time duration) to create unique pulse amplitude for a given combination of event type and its location. In the second option, the harvested event energy is divided into two pulses so that the energy of the first pulse uniquely defines a location and the second pulse uses the remaining energy to identify event types. To minimize classification error at the receiver, we optimize pulse durations in the single-pulse option and pulse energies in the dual-pulse option. Feasibility of eNEUTRAL IoNT is demonstrated using extensive numerical experiments involving terahertz channels. We find that the dual-pulse approach significantly outperforms the single-pulse approach achieving 99\% accuracy for detecting both location and event type in 10-node network monitoring two different event types for a radius of 28 mm. As nanoscale energy harvesters and transmitters are still not available to realize operational event monitoring nodes, we, therefore, evaluate and design a key component of the IoNT

Advancements in nanotechnology promise new capabilities for the Internet of Things (IoT) to monitor extremely fine-grained events with sensors as small as up to a hundred nanometers. Researchers predict that such tiny sensors can be connected to the Internet using graphene-based nano-antenna radiating in the terahertz band, giving rise to the so called Internet of Nano-Things (IoNT). Powering such wireless communications with nanoscale energy supply, however, is a major challenge to overcome. Since in many application domains, different types of events discharge different amounts of energy to the environment, we propose an energy-neutral event monitoring framework, called eNEUTRAL IoNT, that allows the sensors to transmit event information using only the amount of energy harvested from the events. We design and analyze two implementation methods for this framework. The first method uses a single pulse containing the entire energy harvested from the event but manipulates its pulse width (time duration) to create unique pulse amplitude for a given combination of event type and its location. In the second option, the harvested event energy is divided into two pulses so that the energy of the first pulse uniquely defines a location and the second pulse uses the remaining energy to identify event types. To minimize classification error at the receiver, we optimize pulse durations in the single-pulse option and pulse energies in the dual-pulse option. Feasibility of eNEUTRAL IoNT is demonstrated using extensive numerical experiments involving terahertz channels. We find that the dual-pulse approach significantly outperforms the single-pulse approach achieving 99\% accuracy for detecting both location and event type in 10-node network monitoring two different event types for a radius of 28 mm. As nanoscale energy harvesters and transmitters are still not available to realize operational event monitoring nodes, we, therefore, evaluate and design a key component of the IoNT

The evolutions in wireless communication, low-power electronics, and energy harvesting technologies have formed a new computing spectrum widely known as the energy harvesting powered Internet of Things (IoT). In this new vision, IoT devices can harvest energy from the ambient environment to power themselves and can operate for decades without the need for battery recharge. Energy harvesting powered IoT has been widely regarded as a promising candidate in realizing perpetual ubiquitous computing that ensures computing at any time and anywhere. However, the amount of energy that can be harvested is usually unpredictable and limited. Therefore, the high energy consumption in sensing and communication of the IoT devices has become the major impediment to fulfill our vision of permanent operation. This thesis makes four contributions to energy-efficient sensing and communication for kinetic energy harvesting (KEH) powered IoTs. The first contribution is the novel use of KEH signal for human-centric sensing. We propose a KEH-based transportation mode detection system that dramatically reduces the sensing power consumption of conventional system without sacrificing the recognition accuracy. As one step further, the second contribution explores the use of the energy storage component, i.e., capacitor, in kinetic-powered IoT as the sensor to detect human activities. It obviates the need for sampling the continuous motion signal during the activity detection period, whereas, it recognizes the activity by reading the capacitor once every few seconds. The third and fourth contributions of the thesis focus on the design of energy-efficient modulation schemes that enable data sharing between IoT devices using acoustic signal. While current solutions heavily rely on complex and energy-hungry digital signal processing (DSP) to obtain frequency-domain information from the time-domain microphone samples, the third contribution proposes the use of vibration energy harvesting (VEH) as

The rapidly growing demand for higher networking capacity and data rates is forcing researchers to explore the unused spectrum in the higher frequency bands. Two such bands, the millimeter wave (mmWave), ranging from 30 GHz to 300 GHz, and the Terahertz (THz) band, ranging from 0.1 THz to 10 THz, are currently being investigated for possible use in future networks. Because many atmospheric molecules have their natural resonant frequencies in these bands, it is important to understand the effects of molecular absorption and re-radiation on the wireless networking performance in such high frequency bands. Building on the recently discovered molecular absorption models, this thesis conducts a theoretical study on the effect of molecular absorption and re-radiation on both single-antenna and multiantenna wireless communications. For the single-antenna communication, the study focuses on quantifying the temporal and spatial variation of path loss and noise, which is caused by variation in the molecular composition in the air. In particular, it studies the extent of spatio-temporal variation of mmWave channels in three largest cities of Australia by investigating the hourly air quality and weather data over 12 months. The study finds that mmWave channels experience significant variation in both space and time domains, which causes undesirable network capacity fluctuation in various places and hours. For the multi-antenna communication, the study yields a new theoretical discovery that the Multiple-Input and Multiple-Output (MIMO) capacity can be significantly influenced by atmosphere molecules. In more detail, some common atmosphere molecules, such as Oxygen and water, can absorb and re-radiate energy in their natural resonant frequencies, such as 60 GHz, 120GHz and 180 GHz, which belong to the mmWave spectrum. Such phenomenon can provide equivalent Non-Line-of-Sight (NLoS) paths in an environment that is dominated by Line-of-Sight (LoS) transmissions, and thus greatly im

Personal tracking has become an integral part of many context-based applica- tions that rely on user’s current location to deliver their services. Although satellite- based Global Positioning System (GPS) is a popular tracking technology, it is not reliable in indoor and underground tracking scenarios due to the lack of line-of-sight to the satellites. Consequently, smartphone sensors, such as accelerometers and magnetometers, are increasingly being used as a GPS-less tracking solution in such environments.This thesis examines two important GPS-less tracking scenarios, one involving an indoor pedestrian and the other an underground train passenger. For the first case, it considers pedestrian dead reckoning (PDR), which recursively updates the location of the pedestrian by using step length and step heading. A common method to estimate the heading in PDR is to use magnetometer measurements. However, unlike outdoor environments, the Earth’s magnetic field is strongly perturbed inside buildings making magnetometer measurements unreliable for heading estimate. This thesis presents a novel collaborative approach to reduce heading estimation errors when magnetometers are used. The proposed method consists of three components. The first component uses machine learning to detect whether a heading estimate is within a specific error margin. Only heading estimates within the error margin are retained and passed to the second component, which uses data fusion to average the heading estimates from multiple people walking in the same direction. The third component uses signal strengths of radio beacons transmitted by the users to detect the fellow pedestrians walking in the same direction. The rationale of this method is based on the observation that magnetic perturbations are often highly localised in space, and if multiple people are walking in the same direction, then only some of their magnetometers are likely to be perturbed. Extensive evaluations using real data show that

The recent move to empower sensor-rich vehicles with wireless communication technology has created the potential for a city-wide sensing platform that is an essential input to realise a wide range of traffic safety and efficiency applications, that are the building blocks of a future intelligent transportation system (ITS). In particular, many applications, such as adaptive traffic light adjustment, hazard road detection and congestion predictions, are envisioned based on the assumption of the availability of high quality floating car data (FCD). However, efficiently collect- ing and disseminating the huge amount of data generated by a variety of sensors mounted on moving vehicles remains a challenging task due to the limited channel resources, and the highly dynamic and disconnected network topology presented in vehicular networks. In this thesis, we address the above challenges and developed FCD systems in vehicular networks by exploiting the concept of mobile agents. A mobile agent is a self-contained software component that migrates from one host to another, executing application-specific functions. Being highly autonomous, a mobile agent is fundamentally resilient to network faults and offers significant advantages for data collection in highly dynamic and distributed environments. We first address the problem of harvesting FCD from operating vehicles to a remote FCD server, proposing a mobile agent based centralised FCD systems (CFCD) in heterogeneous vehicular networks. The proposed system, inspired by the theory of random walk on graphs, dispatches multiple mobile agents to collect road segment traffic aggregates from vehicles moving within a large urban region. We mathematically prove that given a sufficient number of mobile agents, the system is able to collect all road segments within the region, provides 100% collection coverage in the best case. The proposed CFCD system is further extended to develop a distributed FCD (DFCD) system, such that vehicles c

Recent advancements in drone technology and base station miniaturization, together with an urgent need to reduce site rental costs, have created the unique opportunity to deploy cellular networks on a platform of mobile drones. This new development is redefining the wireless networks, as drone base stations can autonomously move in space to improve coverage and capacity of the network, tremendously enhancing the Quality of Service for conventional cell-edge users.In this research, we explore the benefit of constantly moving drone base stations in the air to reduce the distance between the base stations and the mobile user equipments, thereby improving the performance of the cellular networks.In particular, this thesis makes three fundamental contributions.First, we analyse drone manoeuvrability using theory, emulation and real field experiments to find the relationship between flying speed, turning agility and energy consumption. Under the control of our developed Android program, we reveal some practical manoeuvrability factors that must be considered for the applications that require frequent changes of direction for the drone. Second, we propose drone mobility control algorithms to decide on drones' moving directions in order to improve the performance of drone base stations in the network area. As the optimal problem is NP-hard, we propose a range of practically realizable heuristics with varying complexity and performance. The proposed algorithms are evaluated taking the practical drones' limitations into account for micro hotspots scenario where many hotspots exist next to each other and a drone is deployed over each hotspot area. We show that our proposed heuristic algorithms can readily improve spectral efficiency by 34% and the 5th-percentile packet throughput by 50% compared to the scenario where drones hover over fixed locations. Third, we consider macro hotspot scenario, where users and drones can move freely in a large area. Particular challenges such a

Advances in energy harvesting hardware have created an opportunity to realise self-powered wearables for continuous and pervasive Human Activity Recognition (HAR). Unfortunately, the power requirements of continuous activity sensing using accelerometer sensors and burdensome on-node classification are relatively high compared to the amount of power that can be practically harvested, which limits the usefulness of energy harvesting. This thesis makes three fundamental contributions. First, we propose HARKE, Human Activity Recognition from Kinetic Energy, a novel approach to HAR that does not use an accelerometer. Instead, HARKE employs and infers human physical activities directly from the Kinetic Energy Harvesting (KEH) patterns generated from a device that harvests kinetic energy to power the wearable device. We also show the ability of HARKE to detect related details such as the steps taken by the user in a walking activity. By not using an accelerometer, a significant percentage of the limited harvested energy can be saved. Second, we introduce a novel framework that reduces the on-node classification overhead and guarantees energy neutrality. The proposed framework transmits an unmodulated signal, called an activity pulse, and uses only the received signal strength of the activity pulse to classify human activities. Neither accelerometer nor classifier is required on the wearable device, which therefore, guarantees energy neutrality. Finally, we validate the feasibility of using KEH patterns generated from human speech as a potential new source of information for detecting hotwords, such as ``OK Google", which are used by voice control applications to differentiate user commands from background conversations. Unlike methods that use existing sensors like microphones or accelerometers, our proposal enables pervasive voice control and HAR at a minimum energy cost.We believe that the findings in this thesis will open the door for a new direction of research and dev

Hypertext transfer protocol (HTTP) is the fundamental mechanics supporting web browsing on the Internet. An HTTP server stores large volumes of contents and delivers specific pieces to the clients when requested. There is a recent move to use HTTP for video streaming as well, which promises seamless integration of video delivery to existing HTTP-based server platforms. This is achieved by segmenting the video into many small chunks and storing these chunks as separate files on the server. For adaptive streaming, the server stores different quality versions of the same chunk in different files to allow real-time quality adaptation of the video due to network bandwidth variation experienced by a client. For each chunk of the video, which quality version to download, therefore, becomes a major decision-making challenge for the streaming client, especially in vehicular environment with significant uncertainty in mobile bandwidth. The key objective of this thesis is to explore more advanced decision making tools that would enable an improved tradeoff between conflicting QoE metrics in vehicular environments. In particular, this thesis studies the effectiveness of Markov decision process (MDP), which is known for its ability to optimize decision making under uncertainty. The thesis makes three fundamental contributions: (1) using real video and network bandwidth datasets, it shows that MDP can reduce playback deadline miss of the video (video freezing) by up to 15 times compared to a well known non-MDP strategy when the bandwidth model is known a priori, (2) it proposes a Q-learning implementation of MDP that does not need any a priori knowledge of the bandwidth, but learns optimal decision making in a self-learning manner by simply observing the outcome of its decision making. It is demonstrated that, in terms of deadline miss, the Q-learning-based MDP outperforms the model-based MDP by a factor of three, and (3) it implements the proposed decision making framework in an

Advances in nanotechnology are paving the way for wireless nanoscale sensor networks (WNSNs), promising radically new applications in medical, biological, and chemical fields. However, the small scale poses formidable challenges for communication. First, small nanomaterial-based antennas communicate in the terahertz band, which coincides with the natural resonance frequencies of many types of molecules causing severe molecular absorption and noise. The problem is particularly complicated if the molecular composition of the channel changes over time, causing time-varying absorption and noise. Second, as it is not practical to fit large batteries or replace batteries in a small device, these devices are expected to power themselves by harvesting ambient energy from the environment. However, the amount of energy that can be harvested is directly proportional to the size of the harvester. A nanodevice therefore can generate only a tiny fraction of its total power consumption, which requires us to rethink the design of communication protocols for self-powering WNSNs.In order to address aforementioned challenges, this thesis makes three fundamental contributions. First, it proposes dynamic frequency and power selection as a means to overcome the first problem, i.e, changing molecular composition problem in a time-varying terahertz channel. The dynamic frequency/power selection problem is modelled as a Markov Decision Process to derive the optimal solutions, while several practical heuristics are proposed that achieve close to optimal solutions. Second, to address the severe power shortage problem in a self-powering nanodevice, this thesis proposes a mechanism to exploit the information contained in the energy harvesting data to detect the energy-dissipating events occurring in the environment. This form of event monitoring makes dual use of the energy-harvesting unit in the nanodevice, i.e., it is used to generate power as well as monitor the environment, thus saving sign

Domains such as medicine, the WWW or finance generate on a daily basis massive amounts of data represented as a collection of time sequence of events; Each event is described as a set of descriptors taken from various descriptive domains and associated to a time of occurrence. These archives represent valuable sources of insight for analysts to browse, analyze and discover golden nuggets of knowledge. However, discovering knowledge for such domains is a challenging task since it requires mining massive sequences where the variability of event descriptors could be very high. Recent studies show that data mining methods might need to operate on derived forms of the data including aggregate values or summaries. Knowledge extracted in such a way is called Higher Order Knowledge. In this thesis work, we propose to address this challenge and we define the concept of Time sequence summarization. The purpose is to support applications to scale on very large datasets. Time sequence summarization uses the content and temporal information of events to generate a more concise, yet informative, time sequence that can seamlessly be substituted for the original time sequence in any application. First, we propose a user-oriented technique called TSaR built in a 3-step process: Generalization, grouping and concept formation. TSaR uses taxonomies to generalize event descriptors at higher levels of abstraction. The grouping process then gathers events that are close on the timeline. Second, we reformulate the summarization task as a clustering problem to render it parameter-free. The originality resides on the specificity of the objective function to optimize. Indeed, the objective function considers both the content and the proximity of events on the timeline. We present three solutions including two greedy solutions called G-BUSS and GRASS. Finally, we explore and analyze how summaries contribute to discovering Higher Order Knowledge. We analytically characterize higher order patter

Thanks to the advances in Wireless Wide Area Networks (WWAN) technologies, it is now possible for commuters on moving vehicles to enjoy high-speed Internet connectivity. However, in such high-mobility scenarios, the network conditions (e.g. bandwidth and delay) can fluctuate significantly due to the varying radio conditions along the route. The highly dynamic network environment poses significant difficulties for multimedia applications to maintain Quality of Service (QoS). In this research, we propose a novel framework that addresses this specific problem. Our work is motivated by the possibility of extracting intrinsic patterns of mobile network characteristics from historical observations. For example, there may exist some correlations between network characteristics and geographic location, which can be exploited to forecast the network conditions at the impending location during a vehicular trip. Informed by this a priori knowledge, mobile applications may take proactive decisions to alleviate the impact of network dynamics and maintain QoS while on the move. Our proposed framework consists of three components. The first component involves collecting statistics about actual mobile network conditions under normal driving conditions. From these statistics, the second component is responsible for learning any correlations between network conditions and contextual factors. The last component applies the learned patterns to enable mobile applications to take proactive actions in accommodating network dynamics. In this thesis, we instantiate our framework in the context of downlink WWAN bandwidth. We conduct an extensive vehicular measurement study to collect real-world mobile bandwidth data. Our analysis on the data identifies the dependency between mobile bandwidth and geographic location at granularities as small as 500 metres. We propose bandwidth maps as a means to capture the location-bandwidth dependency, by superimposing the past summary, e.g. the mean, of ba

The concept of motor vehicles interacting with their surroundings using wireless communications is seen as a promising new way to improve road safety and traffic congestion on our increasingly overcrowded roads. While most of the framework has been defined and the necessary wireless spectrum allocated, there still remains many important issues that must be resolved before vehicular communication could realize its full potential. Reliability is one of such issues addressed in this thesis. Packets carrying vital safety information can get lost in harsh and extremely dynamic wireless environment, making reliable communication a challenging problem for vehicles. This thesis proposes novel approaches for packet retransmission techniques to recover from lost packets quickly, effectively, and efficiently in vehicular communication environment. Bandwidth efficiency of the proposed loss recovery techniques is achieved by applying the technique of Network Coding, which allows more lost packets to be recovered per retransmission. Using analytical modeling and computer simulation, the performance of the proposed loss recovery techniques are shown to be significantly better than the previously proposed retransmission techniques. Specifically, it has been shown that Network Coding can achieve higher reliability for the same bandwidth overhead, or higher bandwidth efficiency for the same level of reliability. The packet loss correlation between different receivers has a huge impact on the lost packet recovery. This thesis reports original results from extensive practical experiments to demonstrate that packet losses can be highly uncorrelated even between closely located receivers, providing valuable evidence that Network Coding is a practical concept to pursue for vehicular communications. A final contribution of this thesis is the design and analysis of a dynamic resource scheduling technique that, for a target safety communication reliability, allows efficient integration of bo

Hop count, i.e., the number of wireless hops a packet has to go through to reach the destination, is a fundamental metric in multi-hop wireless ad-hoc networks. Network performance, such as throughput, end-to-end delay, energy consumption, and so on, depends critically on hop count. Previous work on modeling hop count is limited in making unrealistic simplifying assumptions either at the physical or network, or both layers of the communication protocol stack. A key contribution of this thesis is to present an analytical model to derive the probability distribution of hop count under realistic assumptions at both physical and network layers. Specifically, the model considers a log-normal shadowing radio propagation capable of accommodating the random signal fading observed in most wireless communication environments, and the widely used geographic routing at the network layer. Validation of the model is achieved by a comprehensive set of simulation experiments including a trace driven simulation of a real-word vehicular ad-hoc network. The model reveals that the presence of randomness in radio propagation reduces the required number of hops to reach a given destination significantly. To demonstrate the utility of the proposed hop count model, the thesis proposes three new applications which address some of the key challenges in multi-hop wireless networks. The first application derives the per-node packet forwarding load in multi-hop wireless sensor networks and reveals that the nodes in the vicinity of the base station has a significantly less forwarding load than previously thought under simplifying radio propagation and routing assumptions. The second application demonstrates that using hop count as a measure of distance traveled by a data packet, geocasting can be achieved in multi-hop wireless networks in situations when some of the network nodes do not have access to reliable location information. Finally, the proposed hop count model is used to evaluate the pe