Wireless Mesh Networks (WMNs) have emerged as a promising network technology, which combines the benefits of cellular networks and Wireless Local Area Networks (WLANs). In a WMN mesh routers wirelessly relay traffic on behalf of other mesh routers or clients and thereby provide coverage areas comparable to cellular networks, while having the low complexity and low costs of WLANs. As Voice over IP (VoIP) is a very important Internet service, it is critical for the success of WMNs to support high quality VoIP. However, currentWMNs are not adapted well for VoIP. The capacity and scalability of single-radio WMNs is low, especially for small packet transmissions of VoIP calls, because the MAC and PHY layer overhead for small packets is high. The scalability of multiradio/multi-channel WMNs is usually higher, since fewer nodes contend for a channel. However channel scheduling might be required, which can lead to excessive delay and jitter and result in poor VoIP quality. In this thesis we investigate how to deliver high quality VoIP in single radio and multi-radio networks by using cross-layer optimization. For single radio WMNs, we consider the use of IP packet aggregation and IEEE 802.11e transmission opportunities. We conclude that IP packet aggregation greatly improves the capacity and thereby the scalability of WMNs. We show that the key for providing good quality is to artificially delay packets prior to aggregation. We propose a distributed cross-layer optimization system, which, based on Fuzzy Logic Inference, derives an aggregation delay that enhances the capacity and quality. For multi-radio/multi-channel WMNs, we demonstrate the importance of qualityof- service-aware channel scheduling. We develop a quality-of-serviceaware channel scheduler that compared to a basic round-robin scheme significantly reduces jitter and in that way increases VoIP quality. Our analysis shows that there is a trade-off between the jitter of high priority VoIP traffic and the throughput of background TCP traffic. The proposed optimizations significantly increase the capacity of singleradio and multi-radio WMNs. This allows network operators to serve more users with an existing mesh infrastructure or provide better service delivery to existing users.

This thesis presents image processing solutions in FPGA based embedded vision systems. Image processing is a demanding process but the information that can be extracted from images is very useful and can be used for many tasks like mapping and navigation, object detection and recognition, collision detection and more. Image processing or analysis involves reading images from a camera system, improve an image with respect to colour fidelity and white balance, removing distortion, extracting salient information. The mentioned steps are often referred to as low to medium level image processing and involve large amounts of data and fairly simple algorithms suitable for parallel processing. Medium to high level processing involves a reduced amount of data and more complex algorithms. Object recognition which involves matching image features to information stored in a database is of higher complexity. A vision system can be used in anything from a car to industry processes to mobile robots playing soccer or assisting people in their homes. A vision system often works with video streams that are processed to find pieces that can be handled in an industry process, detect obstacles that may be potential hazards in traffic or to find and track landmarks in the environment that can be used to build and navigate from. This involves large amount of calculations and this is a problem, even though modern computers are fast they may still not be able to execute the desired algorithms with the frequency wanted. Even if there are computers that are fast enough they are bulky and require a lot of power. They are not suitable for incorporating on small mobile robots. In this thesis I will present the image processing sequence to give an understanding of the complexity of the processes involved and I will discuss some processing platforms suitable for image processing. I will also present my work that is focused on image algorithm implementations for reconfigurable hardware suitable for mobile robots with requirements on speed an power consumption.

Embedded systems are ubiquitous in the modern world. They are microcomputers most often included incomplete devices consisting of software and hardware. Embedded systems range from small devices to large systems monitoring and controlling complex processes. Design and development of such systems is a complex task, since embedded systems often need to fulfill extra-functional requirements, on top of functional ones, within constrained amounts of platform resources. Some embedded systems are mission critical; hence, they are not allowed to fail during the mission. One way to ensure that a system works in accordance to its specification is to define the system in an Architecture Description Language (ADL) and apply formal verification methods. The Architecture Design and Analysis Language (AADL) has become popular in the avionic and automobile industry, and is equipped with several annexes, among them the Behavior Annex. However, AADL still misses a formal semantics, which prevents the possibility to prove correctness of architecture features by performing model checking on AADL models. Moreover, AADL does not support time annotations, which prevents modeling of real-time systems in AADL. In this thesis, we address these issues by presenting a formal analysis framework including a denotationalsemantics for a subset of the AADL and its Behavior Annex, which evaluates properties defined in Computation Tree Logic (CTL) by providing model checking. Model checking is a formal verification method that has proved to be powerful as well as effective. Our AADL-semantics is supported by a tool with an implementation of the semantics in Standard ML, which in turn is encapsulated in an Eclipse plugin.We also present a time annotation extension of AADL, implemented in a tool translating time annotated AADL and its Behavior Annex into the Timed Abstract State Machine (TASM) for simulation of real-time features. Another closely related problem is how to achieve optimal component distribution; in order to address this issue we have developed a tool that perform near-optional component distribution in regard to a series of parameters. The research results, which have been validated thought case studies, provides the possibility for a system engineer to model a system and prove its correctness. The research has been conducted in the context of the PROGRESS research center, for predictable embedded software systems.

Hierarchical scheduling (also referred to as resource reservation) is a hot topic within the research of real-time systems. It has many advantages such that it can facilitate software integration, fault isolation, structured analysis, legacy system integration etc. The main idea is to partition resources into well defined slots and the resource itself may be the processor, memory etc. This technique is rarely used in real-time applications, however, it is well adopted in the avionics industry in order to isolate error propagation between system parts, and facilitate analysis of the system.Much of the research within resource reservation deals with theoretical schedulability analysis of partitioned systems, including shared resources (other than the processor). We will in this thesis address more practical issues related to resource reservation. We focus on implementation and prototyping aspects, as well as verification and instrumentation. One of our assumptions is that we deal only with fixed-priority preemptive scheduling (FPPS).The first part in this thesis deals with individual software systems that may have its own tasks as well as a scheduler and it is assumed to be part of another larger system, hence, we refer to this individual system as a subsystem. The subsystem is assumed to be integrated together with other subsystems, but at a early stage, we make it possible to simulate the subsystem running together with the rest of the subsystems. This "simulation`` does not require the actual resource reservation mechanism, the only requirement is an operating system with support for FPPS. This pre-study may be a natural step towards the "real`` integration, since each individual subsystem can be test executed within its assigned partition. All subsystems are assumed to run together using a resource reservation mechanism (during the actual integration). We have developed two prototypes of this mechanism. The first prototype is hand-crafted and it is equipped with a program tracer for partitoned based schedulers. This instrumentation is useful for debugging and visualization of program traces for this type of scheduling. The second prototype is developed using timed automata with tasks (task automata). This model-based scheduler is verified for correctness and it is possible to automatically generate source code for the scheduler. We have successfully synthesized this scheduler for the real-time operating system VxWorks. However, it can easily be executed on most other platforms. Both prototypes has pros and cons. The first version has good performance while the second can guarantee its correctness, hence, there is a trade-off between performance and correctness.

The aim of the thesis is object finding in occluded and cluttered environment using computer vision techniques and robot motion. Difficulties of the object finding are 1. finding objects at hidden area and 2. finding unrecognized objects. For solving the difficulties, two methods were developed, one is for finding objects in occluded cluttered environments using model based object finding and the other to increase the robustness in object finding by identifying known objects that are unidentified. The goal was to search occluded areas with the bumblebee2 stereo camera to be able to identify all known objects in the environment by removing all visible known objects To identify known objects SURF [9] was used and to be able to remove the identified objects their location first needed to be localized. To localize the object‘s x and y coordinate the information from SURF [9] was used, and the distance coordinate z is calculated using the depth image from the stereo camera. The method to identify objects the SURF [9] algorithm had missed to identify uses a method to find unknown segments in the environment. By using a push motion on the segments to change their angle it can remove possible light reflections and the object can be identified. The results of this research show that the method can find objects in occluded cluttered areas and it can also identified missed known objects.

No description available.

Education is evolving and is taking advantage of the digital society. Classrooms, study material, communication, etc. are in their traditional senses fading away and being replaced by their digital counterparts. Because of personal computers, faster Internet, and instant information access the need for Web-based education services has appeared. GoLingual is an on-line platform for language learning. It offers education in six languages and supplies all the necessary tools to learn whenever a user chooses to, and by his or her own preferences.In an on-line environment such as GoLingual it seems suitable to lighten up the education and adopt the paradigm of "learning by playing". Therefore it was decided to add a game to the platform. This paper contains an in-depth study about games for education, how they contribute, and what their drawbacks are. Furthermore, it contains a complete walk-through of how the game for the GoLingual platform was disgned. This includes how the game-play was developed, why the design turned out the way it did, and some technical solutions that made the game possible.

In order to perform vulnerability and lethality simulations of vehicles it is necessary to describe their geometry in terms suitable for simulation. AVAL is a piece of software used for such simulations, and a main requirement for input geometry is convexity. Manual decomposition is highly time consuming, and in this thesis, a complete algorithm to perform an automatic convex decomposition is presented and described in more detail than previously done. A proof of concept implementation of the algorithm has been made and tested, and example decompositions are shown. The algorithm is based strongly on earlier work in the eld by Bajaj and Dey, but contains solutions to previously undescribed subproblems. As a result the feasibility of using automatic convex decomposition of vehicle-type geometry as a pre-processing aid is shown.

Coldbase is a column-oriented in-memory database implemented in Java that is used with a specific workload in mind. Coldbase is optimized to receive large streams of timestamped trading data arriving at a fast pace while allowing simple but frequent queries that analyse the data concurrently. By limiting the functionality, Coldbase is able to reach a high performance while the memory consumption is low. This thesis presents ColdbaseTX which is an extension to Coldbase that adds support for transactions. It uses an optimistic approach by storing all writes of a transaction locally and applying them when the transaction commits. Readers are separated from writers by using two versions of the data which makes it possible to guarantee that readers are never blocked.Benchmarks compare Coldbase to ColdbaseTX regarding both performance andmemory efficiency. The results show that ColdbaseTX introduces a small overhead in both memory and performance which however is deemed acceptable since the gain is support for transactions.

Bofors UwS (Underwater Systems) har påbörjat utvecklingen av ett nytt centraliserat system för att koordinera alla torpedhanterare ombord på ett fartyg till en central enhet. Detta system måste kunna ta emot och i vissa fall sända data till de övriga delsystemen ombord på fartyget. För att lyckas med detta behöver de en kommunikationsapplikation till denna centrala enhet. MONOLITH är en prototyp av en sådan applikation. Dess huvudändamål är att demonstrera hur detta kan utföras och att testa enheter som kan kontrollera ett antal torpeder. En sådan enhet kallas för en TIU vilket står för Torpedo Interface Unit. MONOLITH är för närvarande kapabel att kommunicera med ett godtyckligt antal TIU:er (förutsatt att TIU:erna använder sig av TCP/IP-kommunikation) samt ta emot navigationsdata från en GPS-mottagare (Global Positioning System). MONOLITH är även förberett för att implementera flera andra kommunikationsenheter, såsom sonarer, radar o.s.v.