Radio access networks are constantly evolving into a more data intensive direction, emphasizing lower latencies and higher data rates. The growing number of mobile data users and the amount of data they consume requires more data processing capacity from mobile base stations than ever before. As radio access networks evolve according to 3GPP’s plans, so do base station hardware and software. This thesis presents a method for estimating data-link layer processing latencies in an LTE base station. Estimation helps base station manufacturers identify technical performance bottlenecks. It also provides an indication of what level of capacity can be offered to customers. Customers require capacity specifications before products are even implemented so that they can start planning their networks based on these estimated capacity limits. Knowing the capacity limits of new products before they are implemented is a key selling point when the products are being marketed towards network operators. A performance comparison of three different data-link layer implementations of an LTE base station is also presented. Measurement-based worst-case execution time estimation methods are used to create a parameterized model of each implementation. Modeling is done by relating changes in input parameters to changes in the observed execution times using statistical modeling techniques. Measurements are conducted using designed experiments. The resulting models are verified and validated, after which they can be used to estimate processing latencies for different parameter configurations, and to estimate capacity limits of future base station products. / Radioliityntäverkkojen kehityksessä on viime aikoina keskitytty viiveiden lyhentämiseen ja tiedonsiirtonopeuksien kasvattamiseen. Mobiilidatan käyttäjien ja heidän käyttämänsä datan kasvava määrä vaatii tukiasemilta enemmän tiedonkäsittelykapasiteettia kuin koskaan aiemmin. Kun radioliityntäverkkoja kehitetään 3GPP:n kehityssuunnitelmien mukaan, samalla kehittyvät myös verkon tukiasemien laitteistot ja ohjelmistot. Tässä työssä esitetään menetelmä, jonka avulla voidaan ennustaa LTE -tukiaseman siirtoyhteyskerroksen latensseja. Ennustaminen auttaa tukiasemavalmistajia tunnistamaan teknisiä suorituskyvyn pullonkauloja. Ennusteet myös tarjoavat mahdollisuuden arvioida, kuinka paljon kapasiteettia uudet tuotteet voivat tukea. Verkko-operaattorit vaativat tietoa uusien tuotteiden kapasiteettirajoituksista jo ennen kuin tuotteet ovat vielä valmiita. He käyttävät tätä tietoa verkkosuunnittelunsa tukena. Kapasiteettirajoitusten tunteminen jo ennen tuotteiden toteutusta on tärkeä myyntivaltti, kun uusia tukiasematuotteita markkinoidaan operaattoreille. Tässä työssä esitetään myös kolmen erilaisen LTE-tukiaseman siirtoyhteyskerroksen toteutuksen suorituskykyvertailu. Mittauksiin perustuvia huonoimman suoritusajan arviointimenetelmiä käytetään luomaan parametrimalli kustakin toteutuksesta. Mallinnus toteutetaan suhteuttamalla parametrien arvojen muutokset niistä aiheutuvien siirtoyhteyskerroksen käsittelylatenssien muutoksiin käyttäen tilastollisia mallinnusmenetelmiä. Mittaukset toteutetaan käyttäen suunniteltuja kokeita. Tuloksena saatavat mallit todennetaan, minkä jälkeen niitä voidaan käyttää ennustamaan käsittelylatensseja eri parametriyhdistelmille. Malleja voidaan käyttää myös ennustamaan tulevien toteutusten tukemaa kapasiteettia ja suorituskykyä.

Virtual worlds can be potentially anything, ranging from an astronomical simulation to a massively multiplayer online role-playing game. Such a wide variety of potential use cases require a generic and extensible solution when considering the data and communication models for a general-purpose platform for arbitrary virtual worlds with potentially thousands of concurrent users. As computational scalability issues, such as graphics rendering and physics simulation in 3-D environments are already well addressed with existing solutions, this thesis considers the performance problems of multi-user virtual world client-server communication when working on large virtual worlds containing arbitrary data and functionality. Despite the differences in the existing virtual world applications and platforms, all of them share common needs: representing visual 3-D objects, moving and animating them, synchronizing all the necessary data among the participants, and storing it for continuing use and all of this in an efficient manner. In order to succeed in this task, attention must be paid to the scene model that represents the contents of the world, and the protocol that is used to disseminate the changes in the content. Also, by utilizing the available geometric information in the client- server communication and data replication, the data is synchronized between the participants on a need-to-know basis. As a result, this thesis presents a working and production-ready interest manager technique that helps to reduce the server’s bandwidth usage considerably when a world contains lots of moving objects. / Virtuaalimaailmat voivat käsitteenä vaihdella tähtitieteellisistä simulaatiosta massiiviiin monen pelajaan verkottuneisiin roolipeleihen. Jälkimmäisissä tarvitaan geneerinen ja laajennettava alustaratkaisu, jonka tieto- ja kommunikaatiomallit riittävät mielivaltaisille virtuaalimaailmoille, joilla on jopa tuhansia yhtäaikaisia käyttäjiä. Laskennalliset skaalautuvuusongelmat, kuten grafiikan luominen ja fysiikan simulointi kolmiulotteisessa ympäristöissä ovat ovat hyvin tunnettuja. Niistä poiketen tämä diplomityö pohtii suorituskykyongelmia monen käyttäjän virtuaalimaailman asiakas-palvelin-mallin kommunikaatiossa, kun sisältö ja toiminnallisuus voivat olla mielivaltaisia. Olemassa olevien sovellusten ja alustojen eroista huolimatta nillä on yhteisiä tarpeita. Visuaalisia kolmiulotteisia objekteja on esitettävä, niitä on liikutettava ja ja elävöitettävä, sekä jaettava tieto osallistujien kesken synkronoiden. Tämän kaiken on tapahduttava tehokkaasti. Jotta näissä tehtävissä onnistuttaisiin, on kiinnitettävä huomiota maailmamalliin ja protokollaan, jolla välitetään tieto sisällön muutoksista. Hyödyntämällä käytettävissä olevaa geometrista informaatiota asiakas-palvelin-kommunikaatiossa ja tiedon replikoinnissa, tieto synkronoidaan osallistujien kesken vain kun se on oleellista yksittäisen osallistujan kannalta. Tuloksena esitetään toimiva ja käytäntöön soveltuva synkronointialueiden hallintatekniikka. Sen avulla vähennetään palvelimen kaistavaatimuksia merkittävästi, kun maailma sisältää paljon liikkuvia ja muuttuvia kohteita.

As mobile technologies and social media are being integrated more and more to current educational practices, the notion of seamless learning has become important. Seamless learning enables learning across different contexts whenever the students are stimulated to learn. The main motivation behind seamless learning is to encourage the students to switch from one learning context into another easily with the help of a personal device as a mediator. In the most common case this device is a smartphone. One challenge for wider seamless learning adoption is the lack of efficient tools with which teachers could create educational content that is more suitable for seamless learning. To address this challenge, the Interactive Spaces research group at the University of Oulu is implementing an editing tool, called NFC-ACT, with which teachers can create NFC based educational games and exercises for the students. Near Field Communication (NFC) is a short range wireless communication technology in which the devices initiate communication automatically once they are in close proximity. NFC is most commonly used in mobile phones. Thus NFC is a suitable technology for seamless learning applications. In order to test the usability, usefulness and efficiency of NFC-ACT, this thesis presents a detailed usability testing plan for NFC-ACT. With thorough usability testing with teachers as test participants, the most critical usability issues with NFC-ACT can be revealed. In addition, the aim of the usability testing is to study the need for such editing tools. This thesis consists of two parts. The first part is a comprehensive literature review about usability and seamless learning research. Then, the second part presents the current state of the application and goes through in detail the usability testing plan for NFC-ACT. Based on the literature review, the selected methodology for usability testing is an adapted combination of cooperative usability testing and SUXES method. In addition, the testing plan is constructed in a way that it takes into account the results from previously held usability demo sessions that already shed some light on the most critical usability issues with NFC-ACT. The resulted usability testing plan will be put into use once the implementation for NFC-ACT is ready. / Mobiiliteknologiat ja sosiaalinen media ovat integroitumassa yhä enemmän valitseviin opetuskäytäntöihin, minkä myötä saumaton oppiminen on noussut tärkeäksi käsitteeksi. Saumaton oppiminen mahdollistaa oppimisen eri konteksteissa milloin tahansa silloin kun oppilaat ovat stimuloituneita oppimaan. Päämotivaatio saumattoman oppimisen takana on kannustaa oppilaita vaihtamaan oppimiskontekstista toiseen mahdollisimman joustavasti käyttäen hyväksi eri laitteita. Tätä nykyä yleisin tällainen vaihdot mahdollistava laite on älypuhelin. Eräs haaste laajemmalle saumattoman oppimisen omaksumiselle on tehokkaiden työkalujen puute, joiden avulla opettajat voisivat tehdä paremmin saumattomaan oppimiseen soveltuvaa opetusmateriaalia. Oulun yliopiston Interactive Spaces -tutkimusryhmä onkin kehittämässä tähän tarpeeseen editointityökalua nimeltään NFC- ACT, jonka avulla opettajat voivat luoda NFC-pohjasia pelejä ja tehtäviä oppilaille. NFC (Near Field Communication) on lyhyen kantomatkan langaton kommunikointiteknologia, missä laitteet muodostavat automaattisesti yhteyden heti kun ne ovat lähellä toisiaan. Koska NFC:tä käytetään yleisesti matkapuhelimissa, on se sopiva teknologia saumattomaan oppimiseen. Tämä diplomityö esittelee yksityiskohtaisen käytettävyystestaussuunnitelman NFC-ACT:lle, jotta sen käytettävyyttä, hyödyllisyyttä ja tehokkuutta voitaisiin testata. Perusteellisella käytettävyystestauksella opettajien toimiessa testikäyttäjinä työkalun kriittisimmät käytettävyysongelmat voidaan paikallistaa. Lisäksi käytettävyystestauksen tavoitteena on tutkia millainen tarve tällaisille editointi-työkaluille on. Tämä työ koostuu kahdesta osasta. Ensimmäinen osa on kattava selostus käytettävyyteen ja saumattomaan oppimiseen liittyvästä tutkimuksesta. Toinen osa esittelee työkalun nykyisen tilan sekä käy läpi yksityiskohtaisesti käytettävyystestaussuunnitelman. Ensimmäisen osan kirjallisen katsauksen pohjalta käytettävyystestauksen metodiksi muodostui kombinaatio, joka yhdistelee yhteistyökäytettävyystestausta ja SUXES-menetelmiä. Lisäksi testaussuunnitelma ottaa huomioon tulokset NFC-ACT:lle aiemmin järjestetyistä käytettävyysdemoista, jotka jo osaltaan antoivat suuntaa työkalun kriittisimmistä käytettävyysongelmista. Esitelty käytettävyystestaussuunnitelma on tarkoitus ottaa käyttöön heti, kun NFC-ACT:n toteutus on valmis.

Thesis (M.A. (Media studies)) --University of Limpopo, 2016. / Students today are surrounded by vast amounts of technological gadgets that enable communication with friends very easy. Social networking sites, social media and internet have become part of the lives of the students today and the means to their social happiness. This is emphasised by the growing concern over the use of these social networking sites and a growing fear that these sites might actually be causing much damage to the fundamental learning abilities of these students. The purpose of this study was to investigate the impact of the use social networking sites among the students of the University of Limpopo. Qualitative and quantitative research methods were used in this study to explore the traits of the use of social networking sites from the perspective of the students and the lecturers. The data was collected using survey with structured questionnaires for the students and the face-to-face interview with the lecturers to explore their perception over the use of social networking sites. A total of three hundred and nineteen (319) questionnaires were distributed randomly among the students of University of Limpopo and the researcher received two hundred and ninety seven (297) filled questionnaires. Four face-to-face interviews were carried out at the offices of the lecturers that participated in the study. The study revealed that the social networking site Facebook is used more among students and that cell-phones are their means of accessing the internet and social networking sites. The lecturers however pointed out that there is a need for concern with the constant availability of technology at the reach of these students. However, there is a gradual positive development as the students are becoming more and more self-aware that their use of these sites has portends a negative impact on their academic performance. While there are very remote incidents of text language in academic writing contrary to the students’ perception, the lectures believe that student writing quality exists and that what is often perceived as social networking sites language turns out to be innocent spelling errors. Key words Social networking sites, technology, procrastination, distraction, academic performance

Semantic segmentation methods using deep neural networks typically require huge volumes of annotated data to train properly. Due to the expense of collecting these pixel-level dataset annotations, the problem of semantic segmentation without ground-truth labels has been recently proposed. Many current approaches to unsupervised semantic segmentation frame the problem as a pixel clustering task, and in particular focus heavily on color differences between image regions. In this paper, we explore a weakness to this approach: By focusing on color, these approaches do not adequately capture relationships between similar objects across images. We present a new approach to the problem, and propose a novel architecture that captures the characteristic similarities of objects between images directly. We design a synthetic dataset to illustrate this flaw in an existing model. Experiments on this synthetic dataset show that our method can succeed where the pixel color clustering approach fails. Further, we show that plain autoencoder models can implicitly capture these cross-instance object relationships. This suggests that some generative model architectures may be viable candidates for unsupervised semantic segmentation even with no additional loss terms.

Multicast service is one of the methods used to efficiently manage bandwidth when sending multimedia content. To improve bandwidth utilisation, virtualization is often invoked because of its additional features such as bandwidth sharing and support of services that require high volumes of transactional data. Currently, network providers are concerned with the bandwidth amount for efficient use of the limited wireless network capabilities and the provision of a better quality of service. The virtualization design of a multicast service framework should satisfy several objectives. For example, it should enable the interchange of service delivery between multiple networks with one shareable network infrastructure. Also, it should ensure efficient use of network resources and guarantee users' demands of Quality of Service (QoS). Thus, the design of virtualization of multicast service framework is a complex research study. Due to the bandwidth-related arguments, a strong focus has been put on technical issues that facilitate virtualization in wireless networks. A well-designed virtualized network guarantees users with the required quality service. Similarly, virtualization of multicast service is invoked to improve efficient utilisation of bandwidth in wireless networks. As wireless links prove to be unstable, packet loss is unavoidable when multicast service-oriented virtual artefacts are incorporated in wireless networks. In this thesis, a virtualized multicast framework was modelled by using Generalized Assignment Problem (GAP) methodology. Mixed Integer Linear Programing (MILP) was implemented in MATLAB to solve the GAP model. This was to optimise the allocation of multicast traffic to the appropriate virtual networks. Thus, the developed model allows users to have interchangeable services offered by multiple networks. Furthermore, Network Simulator version 3 (NS-3) was used to evaluate the performance of the virtualized multicast framework. Three applications, namely, voice over IP (VoIP), video streaming, and file download have been used to evaluate the performance of a multicast service virtualization framework in Worldwide Interoperability for Microwave Access (WiMAX) networks using NS-3. The performance evaluation was based on whether MILP is used or not used. The results of experimentation have revealed that there is good performance of virtual networks when multicast traffic is sent over one single virtual network instead of sending it over multiple virtual networks. Similarly, the results show that the bandwidth is efficiently used because the multicast traffic is not delivered through multiple virtual networks. Overall, the concepts, the investigations and the model presented in this thesis can enable mobile network providers to achieve efficient use of bandwidth and provide the necessary means to support services for QoS differentiations and guarantees. Also, the multicast service virtualization framework provides an excellent tool that can enable network providers to interchange services. The developed model can serve as a basis for further extension. Specifically, the extension of the model can boost load balancing in the flow allocation problem and activate a virtual network to deliver traffic. This may rely on the QoS policy between network providers. Therefore, the model should consider the number of users in order to guarantee improved QoS.

This study expands current crisis communication research by exploring the communication of instructing information via Twitter. Drawing from the internalization, explanation, and action components of Sellnow & Sellnow's (2013) IDEA Model, this study analyzes live tweets posted by the Boston Police Department during the 2013 Boston Marathon bombing crisis. Examining Tweets posted during a crisis allows us to better understand what constitutes useful, valuable instructing information that can be communicated via social media in real time. Further, scholars have just begun exploring social media's implications for crisis communication. This study extends the IDEA Model to reach crisis communication and social media. Findings also indicate the three components of the IDEA Model are valuable topics to consider when communicating instructing information via Twitter. Finally, we learn that Twitter's 140 character limit does not impede the social media platform as a vehicle to communicate instructing information during a crisis.

archives@tulane.edu / 1 / Jack Green

No description available.

No description available.