Geo-localization is the problem of discovering the location where an image or video was captured. Recently, large scale geo-localization methods which are devised for ground-level imagery and employ techniques similar to image matching have attracted much interest. In these methods, given a reference dataset composed of geo-tagged images, the problem is to estimate the geo-location of a query by finding its matching reference images. In this dissertation, we address three questions central to geo-spatial analysis of ground-level imagery: 1) How to geo-localize images and videos captured at unknown locations? 2) How to refine the geo-location of already geo-tagged data? 3) How to utilize the extracted geo-tags? We present a new framework for geo-locating an image utilizing a novel multiple nearest neighbor feature matching method using Generalized Minimum Clique Graphs (GMCP). First, we extract local features (e.g., SIFT) from the query image and retrieve a number of nearest neighbors for each query feature from the reference data set. Next, we apply our GMCP-based feature matching to select a single nearest neighbor for each query feature such that all matches are globally consistent. Our approach to feature matching is based on the proposition that the first nearest neighbors are not necessarily the best choices for finding correspondences in image matching. Therefore, the proposed method considers multiple reference nearest neighbors as potential matches and selects the correct ones by enforcing the consistency among their global features (e.g., GIST) using GMCP. Our evaluations using a new data set of 102k Street View images shows the proposed method outperforms the state-of-the-art by 10 percent. Geo-localization of images can be extended to geo-localization of a video. We have developed a novel method for estimating the geo-spatial trajectory of a moving camera with unknown intrinsic parameters in a city-scale. The proposed method is based on a three step process: 1) individual geo-localization of video frames using Street View images to obtain the likelihood of the location (latitude and longitude) given the current observation, 2) Bayesian tracking to estimate the frame location and video's temporal evolution using previous state probabilities and current likelihood, and 3) applying a novel Minimum Spanning Trees based trajectory reconstruction to eliminate trajectory loops or noisy estimations. Thus far, we have assumed reliable geo-tags for reference imagery are available through crowdsourcing. However, crowdsourced images are well known to suffer from the acute shortcoming of having inaccurate geo-tags. We have developed the first method for refinement of GPS-tags which automatically discovers the subset of corrupted geo-tags and refines them. We employ Random Walks to discover the uncontaminated subset of location estimations and robustify Random Walks with a novel adaptive damping factor that conforms to the level of noise in the input. In location-aware image understanding, we are interested in improving the image analysis by putting it in the right geo-spatial context. This approach is of particular importance as the majority of cameras and mobile devices are now being equipped with GPS chips. Therefore, developing techniques which can leverage the geo-tags of images for improving the performance of traditional computer vision tasks is of particular interest. We have developed a location-aware multimodal approach which incorporates business directories, textual information, and web images to identify businesses in a geo-tagged query image.

As mobile devices are increasingly merging into our daily lives, exhibition ser- vices are also facing innovation based on the newly available technologies. Our project addresses these new circumstances. We developed a mobile exhibition guide for the exhibition called quot;Mrs Brown's Big Day Out: Hamilton Women in the 1950squot;. That is organized by the Waikato Museum. The proposed sys- tem re-uses the TIP(Tourist Information Provider) system's framework and provides information via mobile devices to visitors on Victoria Street, which is an outdoor part of the exhibition. The information about a sight will be delivered according to visitors' current locations and their interests. We would also like to examine the possibility of re-using our TIP system within the application area of exhibition guide. Therefore, we built the mobile exhibition guide system under the TIP system's framework and developed corresponding services that tailored the system to the requirements of visitors and the exhibition organizer. During the development, we faced a number of challenges, especially, modeling the unknown and unstructured exhibition data into the TIP database. The development process as well as the implementation and evaluation are detailed in this report.

The prevalent visions of ambient intelligence leverage natural interactions between users and available services in a smart space. In recent years, we have seen a huge interest from industry and academia in using handheld devices to interact with things, places and people in the real world. To facilitate such interactions, things are usually annotated with RFID tags or visual markers. These tags or markers are read by a handheld device equipped with an integrated RFID reader or a camera, in order to fetch related information and initiate further actions. Interacting with the Internet of Things (IoT) in a real environment has become increasingly desirable and feasible. This thesis contributes to the domain of physical interactions with IoT; however, we use a spatial-geometric approach instead of RFID or marker based solutions. Using this approach, for example, a user can point his/her handheld device to an annotated thing, from a distance, for the purpose of interaction. The pointing direction and location is determined based on the fusion of the mobile position and of the accelerometer data of the handheld device. To annotate things, their geometric coordinates are specified and related information or services are associated to them. In this thesis, we present a comprehensive and extensible framework to integrate various physical interactions with IoT into multimedia applications. The framework supports the implementations of pointMe, touchMe, and context-aware based interactions with geometrically annotated IoT. We define specific methods and practices that can be incorporated in order to build the interactions. We realize smart home, atlas learning, presentation interaction, smart haptic interaction, and learning based video interaction game prototypes in order to perform experiments and demonstrate the applicability and potential of the proposed geometric based annotation approach. In the analysis of the interaction techniques of the prototypes, we present the advantages and disadvantages of the geometric based annotation of IoT as seen by potential users, in comparison to RFID tags or visual markers based approaches.

The first part of this thesis develops methods for UWB communication. To meet the stringent regulatory body constraints, the physical layer signaling technique of the UWB transceiver should be optimally designed. We propose two signaling schemes which are variants of pulse position modulation (PPM) signaling for impulse radio (IR) UWB communication. We also discuss the detectors for the signaling schemes and evaluate the performance of these detectors.  IR-UWB can be used for precise range measurements as it provides a very high time resolution. This enables accurate time of arrival (TOA) estimations from which precise range values can be derived. We propose methods which use range information to arrive at optimal schedules for an all-to-all broadcast problem. Results indicate that throughput can be increased on average by three to ten times for typical network configurations compared to the traditional methods. Next, we discuss hypothesis testing in the context of UWB transceivers. We show that, when multiple detector outputs from a hardware platform are available, fusing the results from them can yield better performance in hypothesis testing than relying on a single detector output. In the second part of this thesis, the emphasis is placed on localization and joint estimation of location and communication parameters. Here, we focus on estimating the TOA of the signal. The wide bandwidth of the UWB signal requires high speed analog to digital converts (ADC) which makes the cost of the digital transceivers prohibitively high. To address this problem, we take two different strategies. In the first approach, we propose a multichannel receiver with each channel having a low-cost energy detector operating at a sub-Nyquist rate. In the second approach, we consider a compressive sampling based technique. Here, we propose a new acquisition front end, using which the sampling rate of the ADC can be significantly reduced. We extended the idea of compressive sampling based TOA estimation towards joint estimation of TOA and PPM symbols. Here, two signaling methods along with the algorithms are proposed based on the dynamicity of the target. They provide similar performance to the ML based estimation, however with a significant savings in the ADC resources. / <p>QC 20161205</p>

It is predicted that the market for position-based services is going to reach an annual turnover of billions of dollars in a few years. Besides this, also the game industry is estimated to meet a bright future, and there is a great interest already. This master’s thesis looks at a service that combines these two areas. The result is a position-based game called Mystery that is developed for the handheld computer Compaq iPaq 36-serie, which is well suited for this kind of game. To be able to locate a player we have used Ericsson’s Mobile Positioning System, MPS, which uses GSM positioning. The alternative would have been to use the Global Positioning System, GPS, but then the handheld computer would require extra hardware that is not available on the market yet. The prototype consists of a client and a server application. They are both developed in Java and they are communicating with each other through an http- protocol. One conclusion we have drawn form our work is that the kind of game we have developed requires higher accuracy than is achieved with GSM today, even though it is still possible to play the game. The disadvantage now is that the players have to move long distances, which is most likely if they are going by train, bus etc. A better accuracy would make it possible to shrink the distances so that it would also be remunerative to walk. When a better accuracy can be achieved a new possibility for this kind of game will open up. / <p>Validerat; 20101217 (root)</p>

Multi-tier cellular networks have led to a paradigm shift in the deployment of base stations (BSs) where macrocell BSs are overlaid with smaller and lower power BSs such as microcells, picocells, and femtocells. Stochastic geometry has been proven to be an effective tool to capture such heterogeneity and uncertainties in deployment of cellular BSs. In stochastic geometry, random spatial models are used to model multi-tier cellular networks where the locations of BSs is each tier is assumed to be drawn from a point process with the appropriate spatial density. This thesis proposes stochastic geometry-based approaches to analyze, model, and optimize multi-tier cellular networks under several setups and technologies. First, I propose a novel location-aware cross-tier cooperation scheme that aim at improving the performance of users with low signal-to-interference-plus-noise ratio (SINR). Second, I study the performance of cognitive device-to-device (D2D) communication in multi-channel downlink-uplink cellular network with energy harvesting. For the coexistence between cellular and D2D transmissions, I propose a spectrum access policy for cellular BSs to avoid using D2D channels when possible. Third, I investigate the feasibility of energy harvesting from ambient RF interference in multi-tier uplink cellular networks. For this setup, I capture randomness in the network topology and the battery dynamics. Fourth, I extend multi-tier uplink cellular networks to consider the case when users do not necessarily associate with the nearest BS (i.e., flexible cell association). Finally, I compare between different cell association criteria including coupled and decoupled cell association for uplink and downlink transmissions in multi-tier full-duplex cellular networks. For all network setups, I use stochastic geometry to derive simple and closed-form expressions to evaluate the performance in terms of several metrics, e.g., outage probability, mean rate, transmission probability, success probability, and load per BS. I also highlight main tradeoffs in different networks and provide guidelines to optimize different performance metrics by carefully tuning fundamental network design parameters. / February 2017

Εγχειρίδιο χρήσης της Android εφαρμογής WhatWhereWho και παρουσίαση διαδικασίας υλοποίησής της. / Manual for the Android application WhatWhereWho and description of the process needed for its implementation.

No description available.

Commercial provisioning of file synchronization services (FSS) relies entirely on protocols that utilize a remote central server that is often located in the cloud to host important files. Updates at user computers are propagated to the central server and from the central server to other sources in need of such updates. Therefore, a synchronization operation between two computers located on the same network often results in file data transmission to and from this local network at least twice. This introduces an unnecessary bandwidth usage on a user’s local network and has become an issue, especially since there is an increase in the demands for internet resources. This thesis work presents a new file synchronization protocol towards FSS known as LAPSync (location-aware protocol for remote file synchronization). This paper also proposes a hierarchical synchronization mechanism by utilizing LAPSync. Our proposed solution relies on the ability of LAPSync clients to acquire knowledge about the location of clients participating in the synchronization operation to construct a hierarchical synchronization path. Finally, we implement our protocol into a prototype and conduct experiments to compare with drop-box (a popular file synchronization service). The results obtained show that LAPSync offers a reduction in bandwidth usage when the files to be synchronized exist on the same local network