TU München - Fakultät für
Es sprechen Studenten über ihre abgeschlossenen Diplomarbeiten und Systementwicklungsprojekte.
Am Donnerstag, 18.05.17, ab 15:00 Uhr, im Raum Neumann (00.11.038):
Specification and Implementation of Reference Problem for Big Data batch and Stream Analytics
Almost 80% of the world?s data has been generated over the last 2 years and only 10% of this data is being analyzed today. This shows the importance and potential of Big Data in today?s world. This thesis focuses on an important aspect in the domain of Big Data. One of the most important characteristics of Software Engineering is reusability. Whenever we use the term “reusability”, many people believe it to be code reusability. But this is not the case. Requirements, model problems and even their solutions can be reused as part of the same or in a different software project. So far there are not many reusable artifacts in the con- text of Big Data. Our motivation is to prepare such a reusable artifact which can be useful to other developers to start experimenting with the domain of Big Data and extend it to compare different technologies or approaches. This thesis aims to provide the specification and implementation of a reference problem that combines both batch and stream processing in Big Data. The thesis will focus on creating a generic reference problem and then implementing it in a big data cluster using state-of-the-art Big Data tools and technologies. This main goal is broken down into the following goals: 1) Identify a problem that can act as a reference. 2) Specify it in a systematic way (use cases, diagrams). 3) Implement it using a combination of different tools (in order to show that it can act as a pluggable/extensible test-bed for the comparison of different tools) The evaluation is then done on the Big Data tools based on different factors like level of ab- straction, availability of APIs, architecture, maturity of the frameworks/platform, suitable use cases, etc.
Specification and Implementation of Rule Based and Context Aware Smart Home Platform
Internet of Things (IoT) has emerged as one of the most researched topics in the recent years. There have been lots of new technologies in hardware and software that have made it not just a hobby for enthusiasts, but reachable and usable for the common user. An interesting aspect in this research is to identify activities of a user through his interaction with the things around him. These activities can be used as an input to automate management of devices and provide notifications to users, resulting into a smarter and ergonomic environment. For this research, the goal is to setup a context— aware IoT platform that allows receiving information from various devices in a living environment, say a house. This platform will track user’s actions and states of devices through the information received from these devices. The user would be given an interface to explore his IoT devices and create his own contexts. These contexts can then be used to create rules that would trigger actions and notifications. The platform was developed using MVC frameworks and a well known rules engine-Drools. A simple web interface was developed through which the user created rules and contexts. The evaluation of this system was done by a user study and the results of this study were generally positive, with a majority of users agreeing that creating contexts that can be reused in different rules was a good idea.
Cross-context Eventing Mechanism in CDI for JSR 362 Portlet Specification 3.0
Java – A technology which is more than 20 years old now . A language which is con- stantly evolving with time and remains one of the most common denominator for tech giants like Twitter, Google (Android) and Intel . Each new version of Java comes bundled-up with a set of cool new features, changing the way we program our software. Such changes also affect the libraries, Application Programming Interfaces (APIs) or Service-Provider Interfaces (SPIs) which depend on the language. Among thousands of APIs which require change, one such API is the Portlet API. The latest version of Portlet API (Version 3.0) is based on Java enterprise edition 7 and needs to be aligned with the new features included with the language. The most important features that influence the Portlet API are Contexts and Dependency injection (CDI) and support for Java Server Faces (JSF). This research is done with the collaboration of IBM and Technische Universität München. The objective of this research is to enhance the Portlet API 3.0 with the use of CDI events for Inter-Portlet communication which is a communication scheme explained in later part of this document. Currently, it’s possible to use CDI events only within a single web application. In other words, using CDI events is only possible within same application-context. However, by the nature of Inter-Portlet communication scheme, it should be possible to send CDI events across two or more web applications (i.e. in cross-context). This thesis presents a mechanism for sending CDI events in cross-context for Inter-Portlet communication. The research findings propose two ways of sending CDI events in cross-context. First way changes the way CDI events work (even in single web application) and there is no extra con- figuration needed to use this mechanism. The second way proposed an extension of the first. It preserves the way CDI events work in single web application, and devices a new way to send CDI events in cross-context with very little configuration. The result is a very flexible inter- portlet communication scheme which is easier to configure and use than the current scheme. This scheme could become part of the current efforts going on to simplify the programming model of Portlet API and help portlet developers to setup inter-portlet communication in a much quicker and meaningful way.